graphics v0.0.11-depth2

Rather involved update due to it being the foundation of the physics.

Reviewed-on: #45
Co-authored-by: Rhys Lloyd <krakow20@gmail.com>
Co-committed-by: Rhys Lloyd <krakow20@gmail.com>

.cargo/config.toml

@@ -1,6 +1,2 @@
 [registries.strafesnet]
 index = "sparse+https://git.itzana.me/api/packages/strafesnet/cargo/"
-
-[target.x86_64-unknown-linux-gnu]
-linker = "clang"
-rustflags = ["-C", "link-arg=-fuse-ld=/usr/bin/mold"]

.gitignore

@@ -1 +1,2 @@
 /target
+.zed

Cargo.toml

@@ -1,5 +1,10 @@
 [workspace]
 members = [
+	"engine/graphics",
+	"engine/physics",
+	"engine/session",
+	"engine/settings",
+	"integration-testing",
 	"lib/bsp_loader",
 	"lib/common",
 	"lib/deferred_loader",
@@ -7,8 +12,10 @@ members = [
 	"lib/linear_ops",
 	"lib/ratio_ops",
 	"lib/rbx_loader",
+	"lib/rbxassetid",
 	"lib/roblox_emulator",
 	"lib/snf",
+	"map-tool",
 	"strafe-client",
 ]
 resolver = "2"
@@ -17,3 +24,34 @@ resolver = "2"
 #lto = true
 strip = true
 codegen-units = 1
+
+[workspace.lints.rust]
+# unsafe_code = "forbid"
+# missing_docs = "warn"
+# missing_debug_implementations = "warn"
+single_use_lifetimes = "warn"
+trivial_casts = "warn"
+unused_lifetimes = "warn"
+unused_qualifications = "warn"
+# variant_size_differences = "warn"
+unexpected_cfgs = "warn"
+
+[workspace.dependencies]
+glam = "0.32.0"
+wgpu = "29.0.0"
+
+# engine
+strafesnet_graphics = { path = "engine/graphics", registry = "strafesnet" }
+strafesnet_physics = { path = "engine/physics", registry = "strafesnet" }
+strafesnet_session = { path = "engine/session", registry = "strafesnet" }
+strafesnet_settings = { path = "engine/settings", registry = "strafesnet" }
+
+# lib
+fixed_wide = { version = "0.3.0", path = "lib/fixed_wide", registry = "strafesnet" }
+linear_ops = { version = "0.2.0", path = "lib/linear_ops", registry = "strafesnet" }
+ratio_ops = { version = "0.1.0", path = "lib/ratio_ops", registry = "strafesnet" }
+strafesnet_bsp_loader = { version = "0.5.0", path = "lib/bsp_loader", registry = "strafesnet" }
+strafesnet_common = { version = "0.9.0", path = "lib/common", registry = "strafesnet" }
+strafesnet_deferred_loader = { version = "0.6.0", path = "lib/deferred_loader", registry = "strafesnet" }
+strafesnet_rbx_loader = { version = "0.10.2", path = "lib/rbx_loader", registry = "strafesnet" }
+strafesnet_snf = { version = "0.4.0", path = "lib/snf", registry = "strafesnet" }

README.md

@@ -3,6 +3,9 @@
 # Strafe Project
 Monorepo for working on projects related to strafe client.
 
+## Try it out
+See [releases](https://git.itzana.me/StrafesNET/strafe-project/releases) for downloads.
+
 ## How to build and run
 1. Have rust and git installed
 2. `git clone https://git.itzana.me/StrafesNET/strafe-project`
@@ -10,4 +13,4 @@ Monorepo for working on projects related to strafe client.
 4. `cargo run --release --bin strafe-client`
 
 ## Licenses
-Each project has its own license.  Most crates are MIT/Apache but notably the Strafe Client has a sole proprietor license.
+Each project has its own license.  Most crates are MIT/Apache but notably the Strafe Client and engine crates have a sole proprietor license.

engine/graphics/Cargo.toml

@@ -0,0 +1,15 @@
+[package]
+name = "strafesnet_graphics"
+version = "0.0.11-depth2"
+edition = "2024"
+
+[dependencies]
+bytemuck = { version = "1.13.1", features = ["derive"] }
+ddsfile = "0.5.1"
+glam.workspace = true
+id = { version = "0.1.0", registry = "strafesnet" }
+strafesnet_common.workspace = true
+wgpu.workspace = true
+
+[lints]
+workspace = true

engine/graphics/LICENSE

@@ -0,0 +1,8 @@
+/*******************************************************
+* Copyright (C) 2023-2024 Rhys Lloyd <krakow20@gmail.com>
+*
+* This file is part of the StrafesNET bhop/surf client.
+*
+* StrafesNET can not be copied and/or distributed
+* without the express permission of Rhys Lloyd
+*******************************************************/

engine/graphics/shaders/shader.wgsl

@@ -44,11 +44,11 @@ struct ModelInstance{
 	transform:mat4x4<f32>,
 	normal_transform:mat3x3<f32>,
 	color:vec4<f32>,
+	texture_color:vec4<f32>,
 }
 //my fancy idea is to create a megatexture for each model that includes all the textures each intance will need
 //the texture transform then maps the texture coordinates to the location of the specific texture
 //group 1 is the model
-const MAX_MODEL_INSTANCES=512;
 @group(2)
 @binding(0)
 var<uniform> model_instances: array<ModelInstance, MAX_MODEL_INSTANCES>;
@@ -66,6 +66,7 @@ struct EntityOutputTexture {
 	@location(3) view: vec3<f32>,
 	@location(4) color: vec4<f32>,
 	@location(5) @interpolate(flat) model_color: vec4<f32>,
+	@location(6) @interpolate(flat) texture_color: vec4<f32>,
 };
 @vertex
 fn vs_entity_texture(
@@ -81,6 +82,7 @@ fn vs_entity_texture(
 	result.texture = texture;
 	result.color = color;
 	result.model_color = model_instances[instance].color;
+	result.texture_color = model_instances[instance].texture_color;
 	result.view = position.xyz - camera.view_inv[3].xyz;//col(3)
 	result.position = camera.proj * camera.view * position;
 	return result;
@@ -106,7 +108,9 @@ fn fs_entity_texture(vertex: EntityOutputTexture) -> @location(0) vec4<f32> {
 	let d = dot(normal, incident);
 	let reflected = incident - 2.0 * d * normal;
 
-	let fragment_color = textureSample(model_texture, model_sampler, vertex.texture)*vertex.color;
+	let fragment_color = textureSample(model_texture, model_sampler, vertex.texture)*vertex.texture_color;
 	let reflected_color = textureSample(cube_texture, cube_sampler, reflected).rgb;
-	return mix(vec4<f32>(vec3<f32>(0.05) + 0.2 * reflected_color,1.0),mix(vertex.model_color,vec4<f32>(fragment_color.rgb,1.0),fragment_color.a),0.5+0.5*abs(d));
+	let entity_color = mix(vertex.model_color*vertex.color,vec4<f32>(fragment_color.rgb,1.0),fragment_color.a);
+	let sky_color = vec4<f32>(vec3<f32>(0.05) + 0.2 * reflected_color,1.0);
+	return mix(sky_color,entity_color,0.5+0.5*abs(d));
 }

engine/graphics/src/graphics.rs

@@ -3,7 +3,7 @@ use std::collections::{HashSet,HashMap};
 use strafesnet_common::map;
 use strafesnet_common::model::{self, ColorId, NormalId, PolygonIter, PositionId, RenderConfigId, TextureCoordinateId, VertexId};
 use wgpu::{util::DeviceExt,AstcBlock,AstcChannel};
-use crate::model_graphics::{self,IndexedGraphicsMeshOwnedRenderConfig,IndexedGraphicsMeshOwnedRenderConfigId,GraphicsMeshOwnedRenderConfig,GraphicsModelColor4,GraphicsModelOwned,GraphicsVertex};
+use crate::model::{self as model_graphics,IndexedGraphicsMeshOwnedRenderConfig,IndexedGraphicsMeshOwnedRenderConfigId,GraphicsMeshOwnedRenderConfig,GraphicsModelOwned,GraphicsVertex};
 
 struct Indices{
 	count:u32,
@@ -48,6 +48,10 @@ struct GraphicsPipelines{
 	model:wgpu::RenderPipeline,
 }
 
+pub fn view_inv(pos:glam::Vec3,angles:glam::Vec2)->glam::Mat4{
+	//f32 good enough for view matrix
+	glam::Mat4::from_mat3_translation(glam::Mat3::from_euler(glam::EulerRot::YXZ,angles.x,angles.y,0f32),pos)
+}
 struct GraphicsCamera{
 	screen_size:glam::UVec2,
 	fov:glam::Vec2,//slope
@@ -69,15 +73,10 @@ impl GraphicsCamera{
 	pub fn proj(&self)->glam::Mat4{
 		perspective_rh(self.fov.x,self.fov.y,0.4,4000.0)
 	}
-	pub fn world(&self,pos:glam::Vec3,angles:glam::Vec2)->glam::Mat4{
-		//f32 good enough for view matrix
-		glam::Mat4::from_translation(pos)*glam::Mat4::from_euler(glam::EulerRot::YXZ,angles.x,angles.y,0f32)
-	}
 
-	pub fn to_uniform_data(&self,pos:glam::Vec3,angles:glam::Vec2)->[f32;16*4]{
+	pub fn to_uniform_data(&self,view_inv:glam::Mat4)->[f32;16*4]{
 		let proj=self.proj();
 		let proj_inv=proj.inverse();
-		let view_inv=self.world(pos,angles);
 		let view=view_inv.inverse();
 
 		let mut raw=[0f32; 16 * 4];
@@ -88,7 +87,7 @@ impl GraphicsCamera{
 		raw
 	}
 }
-impl std::default::Default for GraphicsCamera{
+impl Default for GraphicsCamera{
 	fn default()->Self{
 		Self{
 			screen_size:glam::UVec2::ONE,
@@ -97,6 +96,28 @@ impl std::default::Default for GraphicsCamera{
 	}
 }
 
+const MODEL_BUFFER_SIZE:usize=4*4 + 12 + 4 + 4;//let size=std::mem::size_of::<ModelInstance>();
+const MODEL_BUFFER_SIZE_BYTES:usize=MODEL_BUFFER_SIZE*4;
+fn get_instances_buffer_data(instances:&[GraphicsModelOwned])->Vec<f32>{
+	let mut raw=Vec::with_capacity(MODEL_BUFFER_SIZE*instances.len());
+	for mi in instances{
+		//model transform
+		raw.extend_from_slice(&AsRef::<[f32; 4*4]>::as_ref(&mi.transform)[..]);
+		//normal transform
+		raw.extend_from_slice(AsRef::<[f32; 3]>::as_ref(&mi.normal_transform.x_axis));
+		raw.extend_from_slice(&[0.0]);
+		raw.extend_from_slice(AsRef::<[f32; 3]>::as_ref(&mi.normal_transform.y_axis));
+		raw.extend_from_slice(&[0.0]);
+		raw.extend_from_slice(AsRef::<[f32; 3]>::as_ref(&mi.normal_transform.z_axis));
+		raw.extend_from_slice(&[0.0]);
+		//color
+		raw.extend_from_slice(AsRef::<[f32; 4]>::as_ref(&mi.color));
+		//texture color
+		raw.extend_from_slice(AsRef::<[f32; 4]>::as_ref(&mi.texture_color));
+	}
+	raw
+}
+
 pub struct GraphicsState{
 	pipelines:GraphicsPipelines,
 	bind_groups:GraphicsBindGroups,
@@ -106,42 +127,45 @@ pub struct GraphicsState{
 	camera_buf:wgpu::Buffer,
 	temp_squid_texture_view:wgpu::TextureView,
 	models:Vec<GraphicsModel>,
+	depth:wgpu::Texture,
 	depth_view:wgpu::TextureView,
 	staging_belt:wgpu::util::StagingBelt,
+	model_instances_uniform_len:usize,
 }
 
 impl GraphicsState{
-	const DEPTH_FORMAT:wgpu::TextureFormat=wgpu::TextureFormat::Depth24Plus;
+	const DEPTH_FORMAT:wgpu::TextureFormat=wgpu::TextureFormat::Depth32Float;
 	fn create_depth_texture(
-		config:&wgpu::SurfaceConfiguration,
+		size:glam::UVec2,
 		device:&wgpu::Device,
-	)->wgpu::TextureView{
-		let depth_texture=device.create_texture(&wgpu::TextureDescriptor{
+	)->wgpu::Texture{
+		device.create_texture(&wgpu::TextureDescriptor{
 			size:wgpu::Extent3d{
-				width:config.width,
-				height:config.height,
+				width:size.x,
+				height:size.y,
 				depth_or_array_layers:1,
 			},
 			mip_level_count:1,
 			sample_count:1,
 			dimension:wgpu::TextureDimension::D2,
 			format:Self::DEPTH_FORMAT,
-			usage:wgpu::TextureUsages::RENDER_ATTACHMENT,
-			label:None,
+			usage:wgpu::TextureUsages::RENDER_ATTACHMENT|wgpu::TextureUsages::COPY_SRC,
+			label:Some("Depth Texture"),
 			view_formats:&[],
-		});
-
-		depth_texture.create_view(&wgpu::TextureViewDescriptor::default())
+		})
+	}
+	pub const fn depth_texture(&self)->&wgpu::Texture{
+		&self.depth
+	}
+	pub const fn depth_texture_view(&self)->&wgpu::TextureView{
+		&self.depth_view
 	}
 	pub fn clear(&mut self){
 		self.models.clear();
 	}
-	pub fn load_user_settings(&mut self,user_settings:&crate::settings::UserSettings){
-		self.camera.fov=user_settings.calculate_fov(1.0,&self.camera.screen_size).as_vec2();
-	}
 	pub fn generate_models(&mut self,device:&wgpu::Device,queue:&wgpu::Queue,map:&map::CompleteMap){
 		//generate texture view per texture
-		let texture_views:HashMap<strafesnet_common::model::TextureId,wgpu::TextureView>=map.textures.iter().enumerate().filter_map(|(texture_id,texture_data)|{
+		let texture_views:HashMap<model::TextureId,wgpu::TextureView>=map.textures.iter().enumerate().filter_map(|(texture_id,texture_data)|{
 			let texture_id=model::TextureId::new(texture_id as u32);
 			let image=match ddsfile::Dds::read(std::io::Cursor::new(texture_data)){
 				Ok(image)=>image,
@@ -213,7 +237,8 @@ impl GraphicsState{
 			let instance=GraphicsModelOwned{
 				transform:model.transform.into(),
 				normal_transform:glam::Mat3::from_cols_array_2d(&model.transform.matrix3.to_array().map(|row|row.map(Into::into))).inverse().transpose(),
-				color:GraphicsModelColor4::new(model.color),
+				color:model.color,
+				texture_color:glam::Vec4::ONE,
 			};
 			//get or create owned mesh map
 			let owned_mesh_map=owned_mesh_id_from_mesh_id_render_config_id
@@ -243,16 +268,15 @@ impl GraphicsState{
 							owned_mesh_id
 						});
 						let owned_mesh=unique_render_config_models.get_mut(owned_mesh_id.get() as usize).unwrap();
-						match &mut owned_mesh.polys{
-							model::PolygonGroup::PolygonList(polygon_list)=>polygon_list.extend(
-								graphics_group.groups.iter().flat_map(|polygon_group_id|{
-									mesh.polygon_groups[polygon_group_id.get() as usize].polys()
-								})
-								.map(|vertex_id_slice|
-									vertex_id_slice.to_vec()
-								)
-							),
-						}
+						let model::PolygonGroup::PolygonList(polygon_list)=&mut owned_mesh.polys;
+						polygon_list.extend(
+							graphics_group.groups.iter().flat_map(|polygon_group_id|{
+								mesh.polygon_groups[polygon_group_id.get() as usize].polys()
+							})
+							.map(|vertex_id_slice|
+								vertex_id_slice.to_vec()
+							)
+						);
 					}
 				}
 				owned_mesh_map
@@ -260,10 +284,20 @@ impl GraphicsState{
 			for owned_mesh_id in owned_mesh_map.values(){
 				let owned_mesh=unique_render_config_models.get_mut(owned_mesh_id.get() as usize).unwrap();
 				let render_config=&map.render_configs[owned_mesh.render_config.get() as usize];
-				if model.color.w==0.0&&render_config.texture.is_none(){
-					continue;
-				}
-				owned_mesh.instances.push(instance.clone());
+				let no_texture=render_config.texture.is_none();
+				let instance_clone=if no_texture{
+					// this model has no chance of affecting pixels on the screen
+					if model.color.w==0.0{
+						continue;
+					}
+					// hack texture_color to zero to make the texture invisible
+					let mut instance=instance.clone();
+					instance.texture_color=glam::Vec4::ZERO;
+					instance
+				}else{
+					instance.clone()
+				};
+				owned_mesh.instances.push(instance_clone);
 			}
 		}
 		//check every model to see if it's using the same (texture,color) but has few instances,if it is combine it into one model
@@ -289,7 +323,7 @@ impl GraphicsState{
 			//separate instances by color
 			for (instance_id,instance) in model.instances.iter().enumerate(){
 				let model_instance_list=unique_color
-				.entry(instance.color)
+				.entry(instance.hashable_color())
 				.or_insert_with(||Vec::new());
 				//add model instance to list
 				model_instance_list.push((model_id,instance_id));
@@ -378,6 +412,7 @@ impl GraphicsState{
 							).collect()
 						));
 					}
+					let (color,texture_color)=color.color();
 					//push model into dedup
 					deduplicated_models.push(IndexedGraphicsMeshOwnedRenderConfig{
 						unique_pos,
@@ -390,7 +425,8 @@ impl GraphicsState{
 						instances:vec![GraphicsModelOwned{
 							transform:glam::Mat4::IDENTITY,
 							normal_transform:glam::Mat3::IDENTITY,
-							color
+							color,
+							texture_color,
 						}],
 					});
 				}
@@ -448,16 +484,14 @@ impl GraphicsState{
 		//.into_iter() the modeldata vec so entities can be /moved/ to models.entities
 		let mut model_count=0;
 		let mut instance_count=0;
-		let uniform_buffer_binding_size=crate::setup::required_limits().max_uniform_buffer_binding_size as usize;
-		let chunk_size=uniform_buffer_binding_size/MODEL_BUFFER_SIZE_BYTES;
 		self.models.reserve(models.len());
 		for model in models.into_iter(){
 			instance_count+=model.instances.len();
-			for instances_chunk in model.instances.rchunks(chunk_size){
+			for instances_chunk in model.instances.rchunks(self.model_instances_uniform_len){
 				model_count+=1;
 				let mut model_uniforms=get_instances_buffer_data(instances_chunk);
 				//TEMP: fill with zeroes to pass validation
-				model_uniforms.resize(MODEL_BUFFER_SIZE*512,0.0f32);
+				model_uniforms.resize(MODEL_BUFFER_SIZE*self.model_instances_uniform_len,0.0f32);
 				let model_buf=device.create_buffer_init(&wgpu::util::BufferInitDescriptor{
 					label:Some(format!("Model{} Buf",model_count).as_str()),
 					contents:bytemuck::cast_slice(&model_uniforms),
@@ -513,7 +547,9 @@ impl GraphicsState{
 	pub fn new(
 		device:&wgpu::Device,
 		queue:&wgpu::Queue,
-		config:&wgpu::SurfaceConfiguration,
+		size:glam::UVec2,
+		view_format:wgpu::TextureFormat,
+		limits:wgpu::Limits,
 	)->Self{
 		let camera_bind_group_layout=device.create_bind_group_layout(&wgpu::BindGroupLayoutDescriptor{
 			label:None,
@@ -590,7 +626,7 @@ impl GraphicsState{
 			address_mode_w:wgpu::AddressMode::ClampToEdge,
 			mag_filter:wgpu::FilterMode::Linear,
 			min_filter:wgpu::FilterMode::Linear,
-			mipmap_filter:wgpu::FilterMode::Linear,
+			mipmap_filter:wgpu::MipmapFilterMode::Linear,
 			..Default::default()
 		});
 		let repeat_sampler=device.create_sampler(&wgpu::SamplerDescriptor{
@@ -600,15 +636,23 @@ impl GraphicsState{
 			address_mode_w:wgpu::AddressMode::Repeat,
 			mag_filter:wgpu::FilterMode::Linear,
 			min_filter:wgpu::FilterMode::Linear,
-			mipmap_filter:wgpu::FilterMode::Linear,
+			mipmap_filter:wgpu::MipmapFilterMode::Linear,
 			anisotropy_clamp:16,
 			..Default::default()
 		});
 
+		let model_instances_uniform_len=limits.max_uniform_buffer_binding_size as usize/MODEL_BUFFER_SIZE_BYTES;
+
+		// write dynamic value into shader
+		let shader=format!("
+			// This is equal to the CHUNK_SIZE constant from graphics.rs
+			const MAX_MODEL_INSTANCES={model_instances_uniform_len};
+		")+include_str!("../shaders/shader.wgsl");
+
 		// Create the render pipeline
 		let shader=device.create_shader_module(wgpu::ShaderModuleDescriptor{
 			label:None,
-			source:wgpu::ShaderSource::Wgsl(Cow::Borrowed(include_str!("shader.wgsl"))),
+			source:wgpu::ShaderSource::Wgsl(Cow::Owned(shader)),
 		});
 
 		//load textures
@@ -724,19 +768,19 @@ impl GraphicsState{
 		let model_pipeline_layout=device.create_pipeline_layout(&wgpu::PipelineLayoutDescriptor{
 			label:None,
 			bind_group_layouts:&[
-				&camera_bind_group_layout,
-				&skybox_texture_bind_group_layout,
-				&model_bind_group_layout,
+				Some(&camera_bind_group_layout),
+				Some(&skybox_texture_bind_group_layout),
+				Some(&model_bind_group_layout),
 			],
-			push_constant_ranges:&[],
+			immediate_size:0,
 		});
 		let sky_pipeline_layout=device.create_pipeline_layout(&wgpu::PipelineLayoutDescriptor{
 			label:None,
 			bind_group_layouts:&[
-				&camera_bind_group_layout,
-				&skybox_texture_bind_group_layout,
+				Some(&camera_bind_group_layout),
+				Some(&skybox_texture_bind_group_layout),
 			],
-			push_constant_ranges:&[],
+			immediate_size:0,
 		});
 
 		// Create the render pipelines
@@ -752,7 +796,7 @@ impl GraphicsState{
 			fragment:Some(wgpu::FragmentState{
 				module:&shader,
 				entry_point:Some("fs_sky"),
-				targets:&[Some(config.view_formats[0].into())],
+				targets:&[Some(view_format.into())],
 				compilation_options:wgpu::PipelineCompilationOptions::default(),
 			}),
 			primitive:wgpu::PrimitiveState{
@@ -761,13 +805,13 @@ impl GraphicsState{
 			},
 			depth_stencil:Some(wgpu::DepthStencilState{
 				format:Self::DEPTH_FORMAT,
-				depth_write_enabled:false,
-				depth_compare:wgpu::CompareFunction::LessEqual,
+				depth_write_enabled:Some(false),
+				depth_compare:Some(wgpu::CompareFunction::LessEqual),
 				stencil:wgpu::StencilState::default(),
 				bias:wgpu::DepthBiasState::default(),
 			}),
 			multisample:wgpu::MultisampleState::default(),
-			multiview:None,
+			multiview_mask:None,
 			cache:None,
 		});
 		let model_pipeline=device.create_render_pipeline(&wgpu::RenderPipelineDescriptor{
@@ -777,7 +821,7 @@ impl GraphicsState{
 				module:&shader,
 				entry_point:Some("vs_entity_texture"),
 				buffers:&[wgpu::VertexBufferLayout{
-					array_stride:std::mem::size_of::<GraphicsVertex>() as wgpu::BufferAddress,
+					array_stride:size_of::<GraphicsVertex>() as wgpu::BufferAddress,
 					step_mode:wgpu::VertexStepMode::Vertex,
 					attributes:&wgpu::vertex_attr_array![0=>Float32x3,1=>Float32x2,2=>Float32x3,3=>Float32x4],
 				}],
@@ -786,7 +830,7 @@ impl GraphicsState{
 			fragment:Some(wgpu::FragmentState{
 				module:&shader,
 				entry_point:Some("fs_entity_texture"),
-				targets:&[Some(config.view_formats[0].into())],
+				targets:&[Some(view_format.into())],
 				compilation_options:wgpu::PipelineCompilationOptions::default(),
 			}),
 			primitive:wgpu::PrimitiveState{
@@ -796,18 +840,18 @@ impl GraphicsState{
 			},
 			depth_stencil:Some(wgpu::DepthStencilState{
 				format:Self::DEPTH_FORMAT,
-				depth_write_enabled:true,
-				depth_compare:wgpu::CompareFunction::LessEqual,
+				depth_write_enabled:Some(true),
+				depth_compare:Some(wgpu::CompareFunction::LessEqual),
 				stencil:wgpu::StencilState::default(),
 				bias:wgpu::DepthBiasState::default(),
 			}),
 			multisample:wgpu::MultisampleState::default(),
-			multiview:None,
+			multiview_mask:None,
 			cache:None,
 		});
 
 		let camera=GraphicsCamera::default();
-		let camera_uniforms=camera.to_uniform_data(glam::Vec3::ZERO,glam::Vec2::ZERO);
+		let camera_uniforms=camera.to_uniform_data(view_inv(glam::Vec3::ZERO,glam::Vec2::ZERO));
 		let camera_buf=device.create_buffer_init(&wgpu::util::BufferInitDescriptor{
 			label:Some("Camera"),
 			contents:bytemuck::cast_slice(&camera_uniforms),
@@ -839,7 +883,8 @@ impl GraphicsState{
 			label:Some("Sky Texture"),
 		});
 
-		let depth_view=Self::create_depth_texture(config,device);
+		let depth=Self::create_depth_texture(size,device);
+		let depth_view=depth.create_view(&wgpu::TextureViewDescriptor::default());
 
 		Self{
 			pipelines:GraphicsPipelines{
@@ -853,46 +898,44 @@ impl GraphicsState{
 			camera,
 			camera_buf,
 			models:Vec::new(),
+			depth,
 			depth_view,
-			staging_belt:wgpu::util::StagingBelt::new(0x100),
+			staging_belt:wgpu::util::StagingBelt::new(device.clone(),0x100),
 			bind_group_layouts:GraphicsBindGroupLayouts{model:model_bind_group_layout},
 			samplers:GraphicsSamplers{repeat:repeat_sampler},
 			temp_squid_texture_view:squid_texture_view,
+			model_instances_uniform_len,
 		}
 	}
 	pub fn resize(
 		&mut self,
 		device:&wgpu::Device,
-		config:&wgpu::SurfaceConfiguration,
-		user_settings:&crate::settings::UserSettings,
+		size:glam::UVec2,
+		fov:glam::Vec2,
 	){
-		self.depth_view=Self::create_depth_texture(config,device);
-		self.camera.screen_size=glam::uvec2(config.width,config.height);
-		self.load_user_settings(user_settings);
+		self.depth=Self::create_depth_texture(size,device);
+		self.depth_view=self.depth.create_view(&wgpu::TextureViewDescriptor::default());
+		self.camera.screen_size=size;
+		self.camera.fov=fov;
 	}
-	pub fn render(
+	pub fn encode_commands(
 		&mut self,
+		encoder:&mut wgpu::CommandEncoder,
 		view:&wgpu::TextureView,
-		device:&wgpu::Device,
-		queue:&wgpu::Queue,
-		frame_state:crate::session::FrameState,
+		camera:glam::Mat4,
 	){
 		//TODO:use scheduled frame times to create beautiful smoothing simulation physics extrapolation assuming no input
 
-		let mut encoder=device.create_command_encoder(&wgpu::CommandEncoderDescriptor{label:None});
-
+		// TODO: find a way to call this directly after queue.submit()
+		self.staging_belt.recall();
 		// update rotation
-		let camera_uniforms=self.camera.to_uniform_data(
-			frame_state.body.extrapolated_position(frame_state.time).map(Into::<f32>::into).to_array().into(),
-			frame_state.camera.simulate_move_angles(glam::IVec2::ZERO)
-		);
+		let camera_uniforms=self.camera.to_uniform_data(camera);
 		self.staging_belt
 			.write_buffer(
-				&mut encoder,
+				encoder,
 				&self.camera_buf,
 				0,
 				wgpu::BufferSize::new((camera_uniforms.len() * 4) as wgpu::BufferAddress).unwrap(),
-				device,
 			)
 			.copy_from_slice(bytemuck::cast_slice(&camera_uniforms));
 		//This code only needs to run when the uniforms change
@@ -927,17 +970,19 @@ impl GraphicsState{
 						}),
 						store:wgpu::StoreOp::Store,
 					},
+					depth_slice:None,
 				})],
 				depth_stencil_attachment:Some(wgpu::RenderPassDepthStencilAttachment{
 					view:&self.depth_view,
 					depth_ops:Some(wgpu::Operations{
 						load:wgpu::LoadOp::Clear(1.0),
-						store:wgpu::StoreOp::Discard,
+						store:wgpu::StoreOp::Store,
 					}),
 					stencil_ops:None,
 				}),
 				timestamp_writes:Default::default(),
 				occlusion_query_set:Default::default(),
+				multiview_mask:None,
 			});
 
 			rpass.set_bind_group(0,&self.bind_groups.camera,&[]);
@@ -955,28 +1000,5 @@ impl GraphicsState{
 			rpass.set_pipeline(&self.pipelines.skybox);
 			rpass.draw(0..3,0..1);
 		}
-
-		queue.submit(std::iter::once(encoder.finish()));
-
-		self.staging_belt.recall();
 	}
 }
-const MODEL_BUFFER_SIZE:usize=4*4 + 12 + 4;//let size=std::mem::size_of::<ModelInstance>();
-const MODEL_BUFFER_SIZE_BYTES:usize=MODEL_BUFFER_SIZE*4;
-fn get_instances_buffer_data(instances:&[GraphicsModelOwned])->Vec<f32>{
-	let mut raw=Vec::with_capacity(MODEL_BUFFER_SIZE*instances.len());
-	for mi in instances{
-		//model transform
-		raw.extend_from_slice(&AsRef::<[f32; 4*4]>::as_ref(&mi.transform)[..]);
-		//normal transform
-		raw.extend_from_slice(AsRef::<[f32; 3]>::as_ref(&mi.normal_transform.x_axis));
-		raw.extend_from_slice(&[0.0]);
-		raw.extend_from_slice(AsRef::<[f32; 3]>::as_ref(&mi.normal_transform.y_axis));
-		raw.extend_from_slice(&[0.0]);
-		raw.extend_from_slice(AsRef::<[f32; 3]>::as_ref(&mi.normal_transform.z_axis));
-		raw.extend_from_slice(&[0.0]);
-		//color
-		raw.extend_from_slice(AsRef::<[f32; 4]>::as_ref(&mi.color.get()));
-	}
-	raw
-}

engine/graphics/src/lib.rs

@@ -0,0 +1,4 @@
+pub mod model;
+pub mod setup;
+pub mod surface;
+pub mod graphics;

engine/graphics/src/model.rs

@@ -30,19 +30,23 @@ pub struct GraphicsMeshOwnedRenderConfig{
 	pub render_config:RenderConfigId,
 	pub instances:Vec<GraphicsModelOwned>,
 }
-#[derive(Clone,Copy,PartialEq,id::Id)]
-pub struct GraphicsModelColor4(glam::Vec4);
-impl std::hash::Hash for GraphicsModelColor4{
-	fn hash<H:std::hash::Hasher>(&self,state:&mut H) {
-		for &f in self.0.as_ref(){
-			bytemuck::cast::<f32,u32>(f).hash(state);
-		}
-	}
-}
-impl Eq for GraphicsModelColor4{}
 #[derive(Clone)]
 pub struct GraphicsModelOwned{
 	pub transform:glam::Mat4,
 	pub normal_transform:glam::Mat3,
-	pub color:GraphicsModelColor4,
+	pub color:glam::Vec4,
+	pub texture_color:glam::Vec4,
+}
+impl GraphicsModelOwned{
+	pub fn hashable_color(&self)->HashableColor{
+		HashableColor([bytemuck::cast::<[f32;4],[u32;4]>(self.color.to_array()),bytemuck::cast::<[f32;4],[u32;4]>(self.texture_color.to_array())])
+	}
+}
+#[derive(Clone,Copy,Eq,Hash,PartialEq)]
+pub struct HashableColor([[u32;4];2]);
+impl HashableColor{
+	pub fn color(self)->(glam::Vec4,glam::Vec4){
+		let [color,texture_color]=bytemuck::cast::<[[u32;4];2],[[f32;4];2]>(self.0);
+		(glam::Vec4::from_array(color),glam::Vec4::from_array(texture_color))
+	}
 }

engine/graphics/src/setup.rs

@@ -0,0 +1,116 @@
+fn optional_features()->wgpu::Features{
+	wgpu::Features::TEXTURE_COMPRESSION_ASTC
+	|wgpu::Features::TEXTURE_COMPRESSION_ETC2
+}
+fn required_features()->wgpu::Features{
+	wgpu::Features::TEXTURE_COMPRESSION_BC
+}
+fn required_downlevel_capabilities()->wgpu::DownlevelCapabilities{
+	wgpu::DownlevelCapabilities{
+		flags:wgpu::DownlevelFlags::empty(),
+		shader_model:wgpu::ShaderModel::Sm5,
+		..wgpu::DownlevelCapabilities::default()
+	}
+}
+
+pub mod step1{
+	pub fn create_instance()->wgpu::Instance{
+		Default::default()
+	}
+}
+
+pub mod step2{
+	pub fn create_surface<'window>(instance:&wgpu::Instance,target:impl Into<wgpu::SurfaceTarget<'window>>)->Result<wgpu::Surface<'window>,wgpu::CreateSurfaceError>{
+		instance.create_surface(target)
+	}
+}
+
+pub mod step3{
+	pub async fn pick_adapter(instance:&wgpu::Instance,surface:&wgpu::Surface<'_>)->Option<wgpu::Adapter>{
+		let backends=wgpu::Backends::from_env().unwrap_or_default();
+		//TODO: prefer adapter that implements optional features
+		//let optional_features=optional_features();
+		let required_features=super::required_features();
+		let required_downlevel_capabilities=super::required_downlevel_capabilities();
+
+		//no helper function smh gotta write it myself
+		let adapters=instance.enumerate_adapters(backends).await;
+
+		let adapter=adapters.into_iter()
+			// reverse because we want to select adapters that appear first in ties,
+			// and max_by_key selects the last equal element in the iterator.
+			.rev()
+			.filter(|adapter|
+				adapter.is_surface_supported(surface)
+				&&adapter.features().contains(required_features)
+				&&{
+					let downlevel_capabilities=adapter.get_downlevel_capabilities();
+					downlevel_capabilities.shader_model>=required_downlevel_capabilities.shader_model
+					&&downlevel_capabilities.flags.contains(required_downlevel_capabilities.flags)
+				}
+			)
+			.max_by_key(|adapter|match adapter.get_info().device_type{
+				wgpu::DeviceType::IntegratedGpu=>3,
+				wgpu::DeviceType::DiscreteGpu=>4,
+				wgpu::DeviceType::VirtualGpu=>2,
+				wgpu::DeviceType::Other|wgpu::DeviceType::Cpu=>1,
+			})?;
+
+		Some(adapter)
+	}
+}
+
+pub mod step4{
+	pub async fn request_device(adapter:&wgpu::Adapter,limits:wgpu::Limits)->Result<(wgpu::Device,wgpu::Queue),wgpu::RequestDeviceError>{
+		let optional_features=super::optional_features();
+		let required_features=super::required_features();
+
+		// Make sure we use the texture resolution limits from the adapter, so we can support images the size of the surface.
+		let needed_limits=limits.using_resolution(adapter.limits());
+
+		let (device, queue)=adapter
+			.request_device(
+				&wgpu::DeviceDescriptor{
+					label:None,
+					required_features:(optional_features&adapter.features())|required_features,
+					required_limits:needed_limits,
+					memory_hints:wgpu::MemoryHints::Performance,
+					trace:wgpu::Trace::Off,
+					experimental_features:wgpu::ExperimentalFeatures::disabled(),
+				},
+			).await?;
+
+		Ok((
+			device,
+			queue,
+		))
+	}
+}
+
+pub mod step5{
+	use crate::surface::Surface;
+	#[derive(Debug)]
+	pub struct ErrorSurfaceNotSupported;
+	impl std::fmt::Display for ErrorSurfaceNotSupported{
+		fn fmt(&self,f:&mut std::fmt::Formatter<'_>)->std::fmt::Result{
+			write!(f,"Surface isn't supported by the adapter.")
+		}
+	}
+	pub fn configure_surface<'window>(
+		adapter:&wgpu::Adapter,
+		device:&wgpu::Device,
+		surface:wgpu::Surface<'window>,
+		(width,height):(u32,u32),
+	)->Result<Surface<'window>,ErrorSurfaceNotSupported>{
+		let mut config=surface
+			.get_default_config(adapter, width, height)
+			.ok_or(ErrorSurfaceNotSupported)?;
+
+		let surface_view_format=config.format.add_srgb_suffix();
+		config.view_formats.push(surface_view_format);
+		config.present_mode=wgpu::PresentMode::AutoNoVsync;
+		surface.configure(device,&config);
+
+		Ok(Surface::new(surface,config))
+	}
+}

engine/graphics/src/surface.rs

@@ -0,0 +1,75 @@
+/// A texture view which can be targeted by draw calls in the command buffer, and then presented to the surface texture.
+pub struct Frame{
+	surface_texture:wgpu::SurfaceTexture,
+	view:wgpu::TextureView,
+}
+impl Frame{
+	pub const fn view(&self)->&wgpu::TextureView{
+		&self.view
+	}
+	pub fn present(self){
+		self.surface_texture.present();
+	}
+}
+
+#[derive(Debug)]
+pub enum FrameError{
+	Skip,
+	DeviceLost,
+}
+impl core::fmt::Display for FrameError{
+	fn fmt(&self,f:&mut std::fmt::Formatter<'_>)->std::fmt::Result{
+		write!(f,"{self:?}")
+	}
+}
+impl core::error::Error for FrameError{}
+
+/// A render surface configuration, containing information such as resolution and pixel format
+pub struct Surface<'window>{
+	surface:wgpu::Surface<'window>,
+	config:wgpu::SurfaceConfiguration,
+}
+
+impl<'window> Surface<'window>{
+	pub(crate) fn new(
+		surface:wgpu::Surface<'window>,
+		config:wgpu::SurfaceConfiguration,
+	)->Self{
+		Self{surface,config}
+	}
+	#[must_use]
+	pub fn new_frame(&self,device:&wgpu::Device)->Result<Frame,FrameError>{
+		let frame=loop{
+			match self.surface.get_current_texture(){
+				wgpu::CurrentSurfaceTexture::Success(surface_texture)=>break surface_texture,
+				// Suboptimal -> surface_texture must be dropped and surface reconfigured
+				wgpu::CurrentSurfaceTexture::Suboptimal(_)
+				|wgpu::CurrentSurfaceTexture::Outdated=>{},
+				wgpu::CurrentSurfaceTexture::Timeout
+				|wgpu::CurrentSurfaceTexture::Occluded=>return Err(FrameError::Skip),
+				wgpu::CurrentSurfaceTexture::Lost=>return Err(FrameError::DeviceLost),
+				wgpu::CurrentSurfaceTexture::Validation=>unreachable!(),
+			};
+			self.surface.configure(device,&self.config);
+		};
+		let view=frame.texture.create_view(&wgpu::TextureViewDescriptor{
+			format:Some(self.config.view_formats[0]),
+			..wgpu::TextureViewDescriptor::default()
+		});
+		Ok(Frame{
+			surface_texture:frame,
+			view,
+		})
+	}
+	pub const fn size(&self)->glam::UVec2{
+		glam::uvec2(self.config.width,self.config.height)
+	}
+	pub fn view_format(&self)->wgpu::TextureFormat{
+		self.config.view_formats[0]
+	}
+	pub fn configure(&mut self,device:&wgpu::Device,size:glam::UVec2){
+		self.config.width=size.x.max(1);
+		self.config.height=size.y.max(1);
+		self.surface.configure(device,&self.config);
+	}
+}

engine/physics/Cargo.toml

@@ -0,0 +1,13 @@
+[package]
+name = "strafesnet_physics"
+version = "0.0.2"
+edition = "2024"
+
+[dependencies]
+arrayvec = "0.7.6"
+glam.workspace = true
+id = { version = "0.1.0", registry = "strafesnet" }
+strafesnet_common.workspace = true
+
+[lints]
+workspace = true

engine/physics/LICENSE

@@ -0,0 +1,8 @@
+/*******************************************************
+* Copyright (C) 2023-2024 Rhys Lloyd <krakow20@gmail.com>
+*
+* This file is part of the StrafesNET bhop/surf client.
+*
+* StrafesNET can not be copied and/or distributed
+* without the express permission of Rhys Lloyd
+*******************************************************/

engine/physics/src/body.rs

@@ -2,12 +2,18 @@ use strafesnet_common::aabb;
 use strafesnet_common::integer::{self,vec3,Time,Planar64,Planar64Vec3};
 #[derive(Clone,Copy,Debug,Hash)]
 pub struct Body<T>{
-	pub position:Planar64Vec3,//I64 where 2^32 = 1 u
-	pub velocity:Planar64Vec3,//I64 where 2^32 = 1 u/s
-	pub acceleration:Planar64Vec3,//I64 where 2^32 = 1 u/s/s
-	pub time:Time<T>,//nanoseconds x xxxxD!
+	pub position:Planar64Vec3,
+	pub velocity:Planar64Vec3,
+	pub time:Time<T>,
 }
-impl<T> std::ops::Neg for Body<T>{
+#[derive(Clone,Copy,Debug,Hash)]
+pub struct Trajectory<T>{
+	pub position:Planar64Vec3,
+	pub velocity:Planar64Vec3,
+	pub acceleration:Planar64Vec3,
+	pub time:Time<T>,
+}
+impl<T> std::ops::Neg for Trajectory<T>{
 	type Output=Self;
 	fn neg(self)->Self::Output{
 		Self{
@@ -18,11 +24,48 @@ impl<T> std::ops::Neg for Body<T>{
 		}
 	}
 }
+impl<T:Copy> std::ops::Neg for &Trajectory<T>{
+	type Output=Trajectory<T>;
+	fn neg(self)->Self::Output{
+		Trajectory{
+			position:self.position,
+			velocity:-self.velocity,
+			acceleration:self.acceleration,
+			time:-self.time,
+		}
+	}
+}
 
 impl<T> Body<T>
 	where Time<T>:Copy,
 {
-	pub const ZERO:Self=Self::new(vec3::ZERO,vec3::ZERO,vec3::ZERO,Time::ZERO);
+	pub const ZERO:Self=Self::new(vec3::zero(),vec3::zero(),Time::ZERO);
+	pub const fn new(position:Planar64Vec3,velocity:Planar64Vec3,time:Time<T>)->Self{
+		Self{
+			position,
+			velocity,
+			time,
+		}
+	}
+	pub const fn with_acceleration(self,acceleration:Planar64Vec3)->Trajectory<T>{
+		let Body{
+			position,
+			velocity,
+			time,
+		}=self;
+		Trajectory{
+			position,
+			velocity,
+			acceleration,
+			time,
+		}
+	}
+}
+
+impl<T> Trajectory<T>
+	where Time<T>:Copy,
+{
+	pub const ZERO:Self=Self::new(vec3::zero(),vec3::zero(),vec3::zero(),Time::ZERO);
 	pub const fn new(position:Planar64Vec3,velocity:Planar64Vec3,acceleration:Planar64Vec3,time:Time<T>)->Self{
 		Self{
 			position,
@@ -31,20 +74,31 @@ impl<T> Body<T>
 			time,
 		}
 	}
+	pub fn relative_to(&self,trj0:&Self,time:Time<T>)->Self{
+		//(p0,v0,a0,t0)
+		//(p1,v1,a1,t1)
+		Trajectory::new(
+			self.extrapolated_position(time)-trj0.extrapolated_position(time),
+			self.extrapolated_velocity(time)-trj0.extrapolated_velocity(time),
+			self.acceleration-trj0.acceleration,
+			time)
+	}
 	pub fn extrapolated_position(&self,time:Time<T>)->Planar64Vec3{
 		let dt=time-self.time;
 		self.position
-		+(self.velocity*dt).map(|elem|elem.divide().fix_1())
-		+self.acceleration.map(|elem|(dt*dt*elem/2).divide().fix_1())
+		+(self.velocity*dt).map(|elem|elem.divide().clamp_64())
+		+self.acceleration.map(|elem|(dt*dt*elem/2).divide().clamp_64())
 	}
 	pub fn extrapolated_velocity(&self,time:Time<T>)->Planar64Vec3{
 		let dt=time-self.time;
-		self.velocity+(self.acceleration*dt).map(|elem|elem.divide().fix_1())
+		self.velocity+(self.acceleration*dt).map(|elem|elem.divide().clamp_64())
 	}
-	pub fn advance_time(&mut self,time:Time<T>){
-		self.position=self.extrapolated_position(time);
-		self.velocity=self.extrapolated_velocity(time);
-		self.time=time;
+	pub fn extrapolated_body(&self,time:Time<T>)->Body<T>{
+		Body::new(
+			self.extrapolated_position(time),
+			self.extrapolated_velocity(time),
+			time,
+		)
 	}
 	pub fn extrapolated_position_ratio_dt<Num,Den,N1,D1,N2,N3,D2,N4,T1>(&self,dt:integer::Ratio<Num,Den>)->Planar64Vec3
 		where
@@ -63,12 +117,12 @@ impl<T> Body<T>
 			D2:Copy,
 			Planar64:core::ops::Mul<D2,Output=N4>,
 			N4:integer::Divide<D2,Output=T1>,
-			T1:integer::Fix<Planar64>,
+			T1:integer::Clamp<Planar64>,
 	{
 		// a*dt^2/2 + v*dt + p
 		// (a*dt/2+v)*dt+p
 		(self.acceleration.map(|elem|dt*elem/2)+self.velocity).map(|elem|dt.mul_ratio(elem))
-		.map(|elem|elem.divide().fix())+self.position
+		.map(|elem|elem.divide().clamp())+self.position
 	}
 	pub fn extrapolated_velocity_ratio_dt<Num,Den,N1,T1>(&self,dt:integer::Ratio<Num,Den>)->Planar64Vec3
 		where
@@ -77,19 +131,21 @@ impl<T> Body<T>
 			Num:core::ops::Mul<Planar64,Output=N1>,
 			Planar64:core::ops::Mul<Den,Output=N1>,
 			N1:integer::Divide<Den,Output=T1>,
-			T1:integer::Fix<Planar64>,
+			T1:integer::Clamp<Planar64>,
 	{
 		// a*dt + v
-		self.acceleration.map(|elem|(dt*elem).divide().fix())+self.velocity
+		self.acceleration.map(|elem|(dt*elem).divide().clamp())+self.velocity
 	}
-	pub fn advance_time_ratio_dt(&mut self,dt:crate::model_physics::GigaTime){
-		self.position=self.extrapolated_position_ratio_dt(dt);
-		self.velocity=self.extrapolated_velocity_ratio_dt(dt);
-		self.time+=dt.into();
+	pub fn extrapolated_body_ratio_dt(&self,dt:crate::model::GigaTime)->Body<T>{
+		Body::new(
+			self.extrapolated_position_ratio_dt(dt),
+			self.extrapolated_velocity_ratio_dt(dt),
+			self.time+dt.into(),
+		)
 	}
 	pub fn infinity_dir(&self)->Option<Planar64Vec3>{
-		if self.velocity==vec3::ZERO{
-			if self.acceleration==vec3::ZERO{
+		if self.velocity==vec3::zero(){
+			if self.acceleration==vec3::zero(){
 				None
 			}else{
 				Some(self.acceleration)
@@ -125,36 +181,12 @@ impl<T> Body<T>
 
 }
 impl<T> std::fmt::Display for Body<T>{
+	fn fmt(&self,f:&mut std::fmt::Formatter<'_>)->std::fmt::Result{
+		write!(f,"p({}) v({}) t({})",self.position,self.velocity,self.time)
+	}
+}
+impl<T> std::fmt::Display for Trajectory<T>{
 	fn fmt(&self,f:&mut std::fmt::Formatter<'_>)->std::fmt::Result{
 		write!(f,"p({}) v({}) a({}) t({})",self.position,self.velocity,self.acceleration,self.time)
 	}
 }
-
-pub struct VirtualBody<'a,T>{
-	body0:&'a Body<T>,
-	body1:&'a Body<T>,
-}
-impl<T> VirtualBody<'_,T>
-	where Time<T>:Copy,
-{
-	pub const fn relative<'a>(body0:&'a Body<T>,body1:&'a Body<T>)->VirtualBody<'a,T>{
-		//(p0,v0,a0,t0)
-		//(p1,v1,a1,t1)
-		VirtualBody{
-			body0,
-			body1,
-		}
-	}
-	pub fn extrapolated_position(&self,time:Time<T>)->Planar64Vec3{
-		self.body1.extrapolated_position(time)-self.body0.extrapolated_position(time)
-	}
-	pub fn extrapolated_velocity(&self,time:Time<T>)->Planar64Vec3{
-		self.body1.extrapolated_velocity(time)-self.body0.extrapolated_velocity(time)
-	}
-	pub fn acceleration(&self)->Planar64Vec3{
-		self.body1.acceleration-self.body0.acceleration
-	}
-	pub fn body(&self,time:Time<T>)->Body<T>{
-		Body::new(self.extrapolated_position(time),self.extrapolated_velocity(time),self.acceleration(),time)
-	}
-}

engine/physics/src/face_crawler.rs

@@ -0,0 +1,187 @@
+use crate::model::{into_giga_time,GigaTime};
+use strafesnet_common::integer::fixed_types::{F64_32,F128_64,F256_128};
+use strafesnet_common::integer::vec3::Vector3;
+use strafesnet_common::integer::{Ratio,Planar64Vec3};
+use crate::physics::{Time,Trajectory};
+use crate::mesh_query::{FEV,DirectedEdge,MeshQuery,MeshTopology};
+
+use core::ops::Bound;
+
+enum Transition<M:MeshTopology>{
+	Miss,
+	Next(FEV<M>,GigaTime),
+	Hit(M::Face,GigaTime),
+}
+
+pub enum CrawlResult<M:MeshTopology>{
+	Miss(FEV<M>),
+	Hit(M::Face,GigaTime),
+}
+impl<M:MeshTopology> CrawlResult<M>{
+	pub fn hit(self)->Option<(M::Face,GigaTime)>{
+		match self{
+			CrawlResult::Miss(_)=>None,
+			CrawlResult::Hit(face,time)=>Some((face,time)),
+		}
+	}
+}
+
+// TODO: move predict_collision_face_out algorithm in here or something
+
+/// check_lower_bound
+pub fn low<LhsNum,LhsDen,RhsNum,RhsDen,T>(lower_bound:&Bound<Ratio<LhsNum,LhsDen>>,dt:&Ratio<RhsNum,RhsDen>)->bool
+where
+	RhsNum:Copy,
+	RhsDen:Copy,
+	LhsNum:Copy,
+	LhsDen:Copy,
+	LhsDen:strafesnet_common::integer::Parity,
+	RhsDen:strafesnet_common::integer::Parity,
+	LhsNum:core::ops::Mul<RhsDen,Output=T>,
+	LhsDen:core::ops::Mul<RhsNum,Output=T>,
+	T:Ord+Copy,
+{
+	match lower_bound{
+		Bound::Included(time)=>time.le_ratio(*dt),
+		Bound::Excluded(time)=>time.lt_ratio(*dt),
+		Bound::Unbounded=>true,
+	}
+}
+/// check_upper_bound
+pub fn upp<LhsNum,LhsDen,RhsNum,RhsDen,T>(dt:&Ratio<LhsNum,LhsDen>,upper_bound:&Bound<Ratio<RhsNum,RhsDen>>)->bool
+where
+	RhsNum:Copy,
+	RhsDen:Copy,
+	LhsNum:Copy,
+	LhsDen:Copy,
+	LhsDen:strafesnet_common::integer::Parity,
+	RhsDen:strafesnet_common::integer::Parity,
+	LhsNum:core::ops::Mul<RhsDen,Output=T>,
+	LhsDen:core::ops::Mul<RhsNum,Output=T>,
+	T:Ord+Copy,
+{
+	match upper_bound{
+		Bound::Included(time)=>dt.le_ratio(*time),
+		Bound::Excluded(time)=>dt.lt_ratio(*time),
+		Bound::Unbounded=>true,
+	}
+}
+
+impl<F:Copy,M:MeshQuery<Normal=Vector3<F>,Offset=F256_128,Position=Planar64Vec3,Direction=Planar64Vec3>> FEV<M>
+	where
+		// This is hardcoded for MinkowskiMesh lol
+		M::Face:Copy,
+		M::Edge:Copy,
+		M::DirectedEdge:Copy,
+		M::Vert:Copy,
+		F:core::ops::Mul<F64_32,Output=F256_128>,
+		<F as core::ops::Mul<F64_32>>::Output:core::iter::Sum,
+		M::Offset:core::ops::Sub<<F as std::ops::Mul<F64_32>>::Output>,
+{
+	fn next_transition(&self,mesh:&M,trajectory:&Trajectory,lower_bound:Bound<GigaTime>,mut upper_bound:Bound<GigaTime>)->Transition<M>{
+		//conflicting derivative means it crosses in the wrong direction.
+		//if the transition time is equal to an already tested transition, do not replace the current best.
+		let mut best_transition=Transition::Miss;
+		match self{
+			&FEV::Face(face_id)=>{
+				//test own face collision time, ignoring roots with zero or conflicting derivative
+				//n=face.normal d=face.dot
+				//n.a t^2+n.v t+n.p-d==0
+				let (n,d)=mesh.face_nd(face_id);
+				//TODO: use higher precision d value?
+				//use the mesh transform translation instead of baking it into the d value.
+				for dt in F256_128::zeroes2((n.dot(trajectory.position)-d)*2,n.dot(trajectory.velocity)*2,n.dot(trajectory.acceleration)){
+					if low(&lower_bound,&dt)&&upp(&dt,&upper_bound)&&n.dot(trajectory.extrapolated_velocity_ratio_dt(dt)).is_negative(){
+						upper_bound=Bound::Included(dt);
+						best_transition=Transition::Hit(face_id,dt);
+						break;
+					}
+				}
+				//test each edge collision time, ignoring roots with zero or conflicting derivative
+				mesh.for_each_face_edge(face_id,|directed_edge_id|{
+					let edge_n=mesh.directed_edge_n(directed_edge_id);
+					let n=n.cross(edge_n);
+					let &[v0,v1]=mesh.edge_verts(directed_edge_id.as_undirected()).as_ref();
+					//WARNING: d is moved out of the *2 block because of adding two vertices!
+					//WARNING: precision is swept under the rug!
+					//wrap for speed
+					for dt in F256_128::zeroes2(n.dot(trajectory.position*2-(mesh.vert(v0)+mesh.vert(v1))).wrap_256(),n.dot(trajectory.velocity).wrap_256()*2,n.dot(trajectory.acceleration).wrap_256()){
+						if low(&lower_bound,&dt)&&upp(&dt,&upper_bound)&&n.dot(trajectory.extrapolated_velocity_ratio_dt(dt)).is_negative(){
+							upper_bound=Bound::Included(dt);
+							best_transition=Transition::Next(FEV::Edge(directed_edge_id.as_undirected()),dt);
+							break;
+						}
+					}
+				});
+				//if none:
+			},
+			&FEV::Edge(edge_id)=>{
+				//test each face collision time, ignoring roots with zero or conflicting derivative
+				let edge_verts=mesh.edge_verts(edge_id);
+				let &[ev0,ev1]=edge_verts.as_ref();
+				let (v0,v1)=(mesh.vert(ev0),mesh.vert(ev1));
+				let edge_n=v1-v0;
+				let delta_pos=trajectory.position*2-(v0+v1);
+				for (i,&edge_face_id) in mesh.edge_faces(edge_id).as_ref().iter().enumerate(){
+					let face_n=mesh.face_nd(edge_face_id).0;
+					//edge_n gets parity from the order of edge_faces
+					let n=face_n.cross(edge_n)*((i as i64)*2-1);
+					//WARNING yada yada d *2
+					//wrap for speed
+					for dt in F256_128::zeroes2(n.dot(delta_pos).wrap_256(),n.dot(trajectory.velocity).wrap_256()*2,n.dot(trajectory.acceleration).wrap_256()){
+						if low(&lower_bound,&dt)&&upp(&dt,&upper_bound)&&n.dot(trajectory.extrapolated_velocity_ratio_dt(dt)).is_negative(){
+							upper_bound=Bound::Included(dt);
+							best_transition=Transition::Next(FEV::Face(edge_face_id),dt);
+							break;
+						}
+					}
+				}
+				//test each vertex collision time, ignoring roots with zero or conflicting derivative
+				for (i,&vert_id) in edge_verts.as_ref().iter().enumerate(){
+					//vertex normal gets parity from vert index
+					let n=edge_n*(1-2*(i as i64));
+					for dt in F128_64::zeroes2((n.dot(trajectory.position-mesh.vert(vert_id)))*2,n.dot(trajectory.velocity)*2,n.dot(trajectory.acceleration)){
+						if low(&lower_bound,&dt)&&upp(&dt,&upper_bound)&&n.dot(trajectory.extrapolated_velocity_ratio_dt(dt)).is_negative(){
+							let dt=Ratio::new(dt.num.widen_256(),dt.den.widen_256());
+							upper_bound=Bound::Included(dt);
+							best_transition=Transition::Next(FEV::Vert(vert_id),dt);
+							break;
+						}
+					}
+				}
+				//if none:
+			},
+			&FEV::Vert(vert_id)=>{
+				//test each edge collision time, ignoring roots with zero or conflicting derivative
+				mesh.for_each_vert_edge(vert_id,|directed_edge_id|{
+					//edge is directed away from vertex, but we want the dot product to turn out negative
+					let n=-mesh.directed_edge_n(directed_edge_id);
+					for dt in F128_64::zeroes2((n.dot(trajectory.position-mesh.vert(vert_id)))*2,n.dot(trajectory.velocity)*2,n.dot(trajectory.acceleration)){
+						if low(&lower_bound,&dt)&&upp(&dt,&upper_bound)&&n.dot(trajectory.extrapolated_velocity_ratio_dt(dt)).is_negative(){
+							let dt=Ratio::new(dt.num.widen_256(),dt.den.widen_256());
+							upper_bound=Bound::Included(dt);
+							best_transition=Transition::Next(FEV::Edge(directed_edge_id.as_undirected()),dt);
+							break;
+						}
+					}
+				});
+				//if none:
+			},
+		}
+		best_transition
+	}
+	pub fn crawl(mut self,mesh:&M,trajectory:&Trajectory,lower_bound:Bound<&Time>,upper_bound:Bound<&Time>)->CrawlResult<M>{
+		let mut lower_bound=lower_bound.map(|&t|into_giga_time(t,trajectory.time));
+		let upper_bound=upper_bound.map(|&t|into_giga_time(t,trajectory.time));
+		for _ in 0..20{
+			match self.next_transition(mesh,trajectory,lower_bound,upper_bound){
+				Transition::Miss=>return CrawlResult::Miss(self),
+				Transition::Next(next_fev,next_time)=>(self,lower_bound)=(next_fev,Bound::Included(next_time)),
+				Transition::Hit(face,time)=>return CrawlResult::Hit(face,time),
+			}
+		}
+		//TODO: fix all bugs
+		//println!("Too many iterations!  Using default behaviour instead of crashing...");
+		CrawlResult::Miss(self)
+	}
+}

engine/physics/src/lib.rs

@@ -0,0 +1,48 @@
+mod body;
+mod face_crawler;
+mod mesh_query;
+mod minkowski;
+mod model;
+mod push_solve;
+mod minimum_difference;
+
+pub mod physics;
+
+// Physics bug fixes can easily desync all bots.
+//
+// When replaying a bot, use the exact physics version which it was recorded with.
+//
+// When validating a new bot, ignore the version and use the latest version,
+// and overwrite the version in the file.
+//
+// Compatible physics versions should be determined
+// empirically at development time via leaderboard resimulation.
+//
+// Compatible physics versions should result in an identical leaderboard state,
+// or the only bots which fail are ones exploiting a surgically patched bug.
+#[derive(Clone,Copy,Hash,Debug,id::Id,Eq,PartialEq,Ord,PartialOrd)]
+pub struct PhysicsVersion(u32);
+pub const VERSION:PhysicsVersion=PhysicsVersion(0);
+const LATEST_COMPATIBLE_VERSION:[u32;1+VERSION.0 as usize]=const{
+	let compat=[0];
+
+	let mut input_version=0;
+	while input_version<compat.len(){
+		// compatible version must be greater than or equal to the input version
+		assert!(input_version as u32<=compat[input_version]);
+		// compatible version must be a version that exists
+		assert!(compat[input_version]<=VERSION.0);
+		input_version+=1;
+	}
+	compat
+};
+pub enum PhysicsVersionError{
+	UnknownPhysicsVersion,
+}
+pub const fn get_latest_compatible_version(PhysicsVersion(version):PhysicsVersion)->Result<PhysicsVersion,PhysicsVersionError>{
+	if (version as usize)<LATEST_COMPATIBLE_VERSION.len(){
+		Ok(PhysicsVersion(LATEST_COMPATIBLE_VERSION[version as usize]))
+	}else{
+		Err(PhysicsVersionError::UnknownPhysicsVersion)
+	}
+}

engine/physics/src/mesh_query.rs

@@ -0,0 +1,57 @@
+#[derive(Debug)]
+pub enum FEV<M:MeshTopology>{
+	Vert(M::Vert),
+	Edge(M::Edge),
+	Face(M::Face),
+}
+
+pub trait UndirectedEdge{
+	type DirectedEdge:DirectedEdge<UndirectedEdge=Self>;
+	fn as_directed(self,parity:bool)->Self::DirectedEdge;
+}
+pub trait DirectedEdge{
+	type UndirectedEdge:UndirectedEdge<DirectedEdge=Self>;
+	fn as_undirected(self)->Self::UndirectedEdge;
+	fn parity(&self)->bool;
+	fn reverse(self)->Self
+		where
+			Self:Sized
+	{
+		let parity=!self.parity();
+		self.as_undirected().as_directed(parity)
+	}
+}
+
+pub trait MeshTopology{
+	type Face;
+	type Edge:UndirectedEdge<DirectedEdge=Self::DirectedEdge>;
+	type DirectedEdge:DirectedEdge<UndirectedEdge=Self::Edge>;
+	type Vert;
+	fn for_each_vert_edge(&self,vert_id:Self::Vert,f:impl FnMut(Self::DirectedEdge));
+	fn for_each_vert_face(&self,vert_id:Self::Vert,f:impl FnMut(Self::Face));
+	fn edge_faces(&self,edge_id:Self::Edge)->impl AsRef<[Self::Face;2]>;
+	fn edge_verts(&self,edge_id:Self::Edge)->impl AsRef<[Self::Vert;2]>;
+	#[expect(unused)]
+	fn for_each_face_vert(&self,face_id:Self::Face,f:impl FnMut(Self::Vert));
+	fn for_each_face_edge(&self,face_id:Self::Face,f:impl FnMut(Self::DirectedEdge));
+}
+
+// Make face_nd d value relative
+// euclidean point?
+// Simplex physics
+// Directed edge necessary?
+// recursive for_each function calls
+// define faces from vertices (Fixed<2> vs Fixed<3>)
+pub trait MeshQuery:MeshTopology{
+	type Position;
+	type Direction;
+	type Normal;
+	type Offset;
+	fn vert(&self,vert_id:Self::Vert)->Self::Position;
+	fn farthest_vert(&self,dir:Self::Direction)->Self::Vert;
+	/// This must return a point inside the mesh.
+	fn hint_point(&self)->Self::Position;
+	fn face_nd(&self,face_id:Self::Face)->(Self::Normal,Self::Offset);
+	fn edge_n(&self,edge_id:Self::Edge)->Self::Direction;
+	fn directed_edge_n(&self,directed_edge_id:Self::DirectedEdge)->Self::Direction;
+}

engine/physics/src/minimum_difference.rs

@@ -0,0 +1,924 @@
+use strafesnet_common::integer::fixed_types::{F128_64,F192_96,F256_128};
+use strafesnet_common::integer::vec3;
+use strafesnet_common::integer::vec3::Vector3;
+use strafesnet_common::integer::{Fixed,Planar64,Planar64Vec3};
+
+use crate::mesh_query::{FEV,DirectedEdge,MeshQuery,MeshTopology};
+// TODO: remove mesh invert
+use crate::minkowski::{MinkowskiMesh,MinkowskiVert};
+
+// This algorithm is based on Lua code
+// written by Trey Reynolds in 2021
+
+type Simplex<const N:usize,Vert>=[Vert;N];
+#[derive(Clone,Copy)]
+enum Simplex1_3<Vert>{
+	Simplex1(Simplex<1,Vert>),
+	Simplex2(Simplex<2,Vert>),
+	Simplex3(Simplex<3,Vert>),
+}
+impl<Vert> Simplex1_3<Vert>{
+	fn push_front(self,v:Vert)->Simplex2_4<Vert>{
+		match self{
+			Simplex1_3::Simplex1([v0])=>Simplex2_4::Simplex2([v,v0]),
+			Simplex1_3::Simplex2([v0,v1])=>Simplex2_4::Simplex3([v,v0,v1]),
+			Simplex1_3::Simplex3([v0,v1,v2])=>Simplex2_4::Simplex4([v,v0,v1,v2]),
+		}
+	}
+}
+#[derive(Clone,Copy)]
+enum Simplex2_4<Vert>{
+	Simplex2(Simplex<2,Vert>),
+	Simplex3(Simplex<3,Vert>),
+	Simplex4(Simplex<4,Vert>),
+}
+
+/*
+local function absDet(r, u, v, w)
+	if w then
+		return math.abs((u - r):Cross(v - r):Dot(w - r))
+	elseif v then
+		return (u - r):Cross(v - r).magnitude
+	elseif u then
+		return (u - r).magnitude
+	else
+		return 1
+	end
+end
+*/
+impl<Vert> Simplex2_4<Vert>{
+	fn det_is_zero<M:MeshQuery<Vert=Vert,Position=Planar64Vec3>>(self,mesh:&M)->bool{
+		match self{
+			Self::Simplex4([p0,p1,p2,p3])=>{
+				let p0=mesh.vert(p0);
+				let p1=mesh.vert(p1);
+				let p2=mesh.vert(p2);
+				let p3=mesh.vert(p3);
+				(p1-p0).cross(p2-p0).dot(p3-p0)==Fixed::ZERO
+			},
+			Self::Simplex3([p0,p1,p2])=>{
+				let p0=mesh.vert(p0);
+				let p1=mesh.vert(p1);
+				let p2=mesh.vert(p2);
+				(p1-p0).cross(p2-p0)==vec3::zero()
+			},
+			Self::Simplex2([p0,p1])=>{
+				let p0=mesh.vert(p0);
+				let p1=mesh.vert(p1);
+				p1-p0==vec3::zero()
+			}
+		}
+	}
+}
+
+/*
+local function choosePerpendicularDirection(d)
+	local x, y, z = d.x, d.y, d.z
+	local best = math.min(x*x, y*y, z*z)
+	if x*x == best then
+		return Vector3.new(y*y + z*z, -x*y, -x*z)
+	elseif y*y == best then
+		return Vector3.new(-x*y, x*x + z*z, -y*z)
+	else
+		return Vector3.new(-x*z, -y*z, x*x + y*y)
+	end
+end
+*/
+fn choose_perpendicular_direction(d:Planar64Vec3)->Planar64Vec3{
+	let x=d.x.abs();
+	let y=d.y.abs();
+	let z=d.z.abs();
+	if x<y&&x<z{
+		Vector3::new([Fixed::ZERO,-d.z,d.y])
+	}else if y<z{
+		Vector3::new([d.z,Fixed::ZERO,-d.x])
+	}else{
+		Vector3::new([-d.y,d.x,Fixed::ZERO])
+	}
+}
+
+const fn choose_any_direction()->Planar64Vec3{
+	vec3::X
+}
+
+fn narrow_dir2(dir:Vector3<F128_64>)->Planar64Vec3{
+	if dir==vec3::zero(){
+		return dir.narrow_64().unwrap();
+	}
+	let x=dir.x.as_bits().unsigned_abs().bit_width();
+	let y=dir.y.as_bits().unsigned_abs().bit_width();
+	let z=dir.z.as_bits().unsigned_abs().bit_width();
+	let big=x.max(y).max(z);
+	const MAX_BITS:u32=64+31;
+	if MAX_BITS<big{
+		dir>>(big-MAX_BITS)
+	}else{
+		dir
+	}.narrow_64().unwrap()
+}
+fn narrow_dir3(dir:Vector3<F192_96>)->Planar64Vec3{
+	if dir==vec3::zero(){
+		return dir.narrow_64().unwrap();
+	}
+	let x=dir.x.as_bits().unsigned_abs().bit_width();
+	let y=dir.y.as_bits().unsigned_abs().bit_width();
+	let z=dir.z.as_bits().unsigned_abs().bit_width();
+	let big=x.max(y).max(z);
+	const MAX_BITS:u32=96+31;
+	if MAX_BITS<big{
+		dir>>(big-MAX_BITS)
+	}else{
+		dir
+	}.narrow_64().unwrap()
+}
+
+fn reduce1<M:MeshQuery<Position=Planar64Vec3>>(
+	[v0]:Simplex<1,M::Vert>,
+	mesh:&M,
+	point:Planar64Vec3,
+)->Reduced<M::Vert>
+	where M::Vert:Copy,
+{
+	// --debug.profilebegin("reduceSimplex0")
+	// local a = a1 - a0
+	let p0=mesh.vert(v0);
+
+	// local p = -a
+	let p=-(p0+point);
+
+	// local direction = p
+	let mut dir=p;
+
+	// if direction.magnitude == 0 then
+	// 	direction = chooseAnyDirection()
+	if dir==vec3::zero(){
+		dir=choose_any_direction();
+	}
+
+	// return direction, a0, a1
+	Reduced{
+		dir,
+		simplex:Simplex1_3::Simplex1([v0]),
+	}
+}
+
+// local function reduceSimplex1(a0, a1, b0, b1)
+fn reduce2<M:MeshQuery<Position=Planar64Vec3>>(
+	[v0,v1]:Simplex<2,M::Vert>,
+	mesh:&M,
+	point:Planar64Vec3,
+)->Reduced<M::Vert>
+	where
+		M::Vert:Copy
+{
+	// --debug.profilebegin("reduceSimplex1")
+	// local a = a1 - a0
+	// local b = b1 - b0
+	let p0=mesh.vert(v0);
+	let p1=mesh.vert(v1);
+
+	// local p = -a
+	// local u = b - a
+	let p=-(p0+point);
+	let u=p1-p0;
+
+	// -- modify to take into account the radiuses
+	// local p_u = p:Dot(u)
+	let p_u=p.dot(u);
+
+	// if p_u >= 0 then
+	if !p_u.is_negative(){
+		// local direction = u:Cross(p):Cross(u)
+		let direction=u.cross(p).cross(u);
+
+		// if direction.magnitude == 0 then
+		if direction==vec3::zero(){
+			return Reduced{
+				dir:choose_perpendicular_direction(u),
+				simplex:Simplex1_3::Simplex2([v0,v1]),
+			};
+		}
+
+		// -- modify the direction to take into account a0R and b0R
+		// return direction, a0, a1, b0, b1
+		return Reduced{
+			dir:narrow_dir3(direction),
+			simplex:Simplex1_3::Simplex2([v0,v1]),
+		};
+	}
+
+	// local direction = p
+	let mut dir=p;
+
+	// if direction.magnitude == 0 then
+	if dir==vec3::zero(){
+		dir=choose_perpendicular_direction(u);
+	}
+
+	// return direction, a0, a1
+	Reduced{
+		dir,
+		simplex:Simplex1_3::Simplex1([v0]),
+	}
+}
+
+// local function reduceSimplex2(a0, a1, b0, b1, c0, c1)
+fn reduce3<M:MeshQuery<Position=Planar64Vec3>>(
+	[v0,mut v1,v2]:Simplex<3,M::Vert>,
+	mesh:&M,
+	point:Planar64Vec3,
+)->Reduced<M::Vert>
+	where
+		M::Vert:Copy
+{
+	// --debug.profilebegin("reduceSimplex2")
+	// local a = a1 - a0
+	// local b = b1 - b0
+	// local c = c1 - c0
+	let p0=mesh.vert(v0);
+	let p1=mesh.vert(v1);
+	let p2=mesh.vert(v2);
+
+	// local p = -a
+	// local u = b - a
+	// local v = c - a
+	let p=-(p0+point);
+	let mut u=p1-p0;
+	let v=p2-p0;
+
+	// local uv = u:Cross(v)
+	// local up = u:Cross(p)
+	// local pv = p:Cross(v)
+	// local uv_up = uv:Dot(up)
+	// local uv_pv = uv:Dot(pv)
+	let mut uv=u.cross(v);
+	let mut up=u.cross(p);
+	let pv=p.cross(v);
+	let uv_up=uv.dot(up);
+	let uv_pv=uv.dot(pv);
+
+	// if uv_up >= 0 and uv_pv >= 0 then
+	if !uv_up.is_negative()&&!uv_pv.is_negative(){
+		// local uvp = uv:Dot(p)
+		let uvp=uv.dot(p);
+
+		// local direction = uvp < 0 and -uv or uv
+		let direction=if uvp.is_negative(){
+			-uv
+		}else{
+			uv
+		};
+
+		// return direction, a0, a1, b0, b1, c0, c1
+		return Reduced{
+			dir:narrow_dir2(direction),
+			simplex:Simplex1_3::Simplex3([v0,v1,v2]),
+		};
+	}
+
+	// local u_u = u:Dot(u)
+	// local v_v = v:Dot(v)
+	// local uDist = uv_up/(u_u*v.magnitude)
+	// local vDist = uv_pv/(v_v*u.magnitude)
+	// local minDist2 = math.min(uDist, vDist)
+	let u_dist=uv_up*v.length();
+	let v_dist=uv_pv*u.length();
+
+	// if vDist == minDist2 then
+	if v_dist<u_dist{
+		u=v;
+		up=-pv;
+		uv=-uv;
+		// b0 = c0
+		// b1 = c1
+		v1=v2;
+	}
+
+	// local p_u = p:Dot(u)
+	let p_u=p.dot(u);
+
+	// if p_u >= 0 then
+	if !p_u.is_negative(){
+		// local direction = up:Cross(u)
+		let direction=up.cross(u);
+		// if direction.magnitude == 0 then
+		if direction==vec3::zero(){
+			// direction = uv
+			return Reduced{
+				dir:narrow_dir2(uv),
+				simplex:Simplex1_3::Simplex2([v0,v1]),
+			};
+		}
+
+		// return direction, a0, a1, b0, b1
+		return Reduced{
+			dir:narrow_dir3(direction),
+			simplex:Simplex1_3::Simplex2([v0,v1]),
+		};
+	}
+
+	// local direction = p
+	let dir=p;
+	// if direction.magnitude == 0 then
+	if dir==vec3::zero(){
+		// direction = uv
+		return Reduced{
+			dir:narrow_dir2(uv),
+			simplex:Simplex1_3::Simplex1([v0]),
+		};
+	}
+	// return direction, a0, a0
+	Reduced{
+		dir,
+		simplex:Simplex1_3::Simplex1([v0]),
+	}
+}
+
+// local function reduceSimplex3(a0, a1, b0, b1, c0, c1, d0, d1)
+fn reduce4<M:MeshQuery<Position=Planar64Vec3>>(
+	[v0,mut v1,mut v2,v3]:Simplex<4,M::Vert>,
+	mesh:&M,
+	point:Planar64Vec3,
+)->Reduce<M::Vert>
+	where
+		M::Vert:Copy
+{
+	// --debug.profilebegin("reduceSimplex3")
+	// local a = a1 - a0
+	// local b = b1 - b0
+	// local c = c1 - c0
+	// local d = d1 - d0
+	let p0=mesh.vert(v0);
+	let p1=mesh.vert(v1);
+	let p2=mesh.vert(v2);
+	let p3=mesh.vert(v3);
+
+	// local p = -a
+	// local u = b - a
+	// local v = c - a
+	// local w = d - a
+	let p=-(p0+point);
+	let mut u=p1-p0;
+	let mut v=p2-p0;
+	let w=p3-p0;
+
+	// local uv = u:Cross(v)
+	// local vw = v:Cross(w)
+	// local wu = w:Cross(u)
+	// local uvw = uv:Dot(w)
+	// local pvw = vw:Dot(p)
+	// local upw = wu:Dot(p)
+	// local uvp = uv:Dot(p)
+	let mut uv=u.cross(v);
+	let vw=v.cross(w);
+	let wu=w.cross(u);
+	let uv_w=uv.dot(w);
+	let pv_w=vw.dot(p);
+	let up_w=wu.dot(p);
+	let uv_p=uv.dot(p);
+
+	// if pvw/uvw >= 0 and upw/uvw >= 0 and uvp/uvw >= 0 then
+	if !pv_w.div_sign(uv_w).is_negative()
+	 &&!up_w.div_sign(uv_w).is_negative()
+	 &&!uv_p.div_sign(uv_w).is_negative(){
+		// origin is contained, this is a positive detection
+		// local direction = Vector3.new(0, 0, 0)
+		// return direction, a0, a1, b0, b1, c0, c1, d0, d1
+		return Reduce::Escape([v0,v1,v2,v3]);
+	}
+
+	// local uvwSign = uvw < 0 and -1 or uvw > 0 and 1 or 0
+	// local uvDist = uvp*uvwSign/uv.magnitude
+	// local vwDist = pvw*uvwSign/vw.magnitude
+	// local wuDist = upw*uvwSign/wu.magnitude
+	// local minDist3 = math.min(uvDist, vwDist, wuDist)
+	let uv_dist=uv_p.mul_sign(uv_w);
+	let vw_dist=pv_w.mul_sign(uv_w);
+	let wu_dist=up_w.mul_sign(uv_w);
+	let wu_len=wu.length();
+	let uv_len=uv.length();
+	let vw_len=vw.length();
+
+	if vw_dist*wu_len<wu_dist*vw_len{
+		// if vwDist == minDist3 then
+		if vw_dist*uv_len<uv_dist*vw_len{
+			(u,v)=(v,w);
+			uv=vw;
+			// uv_p=pv_w; // unused
+			// b0, c0 = c0, d0
+			// b1, c1 = c1, d1
+			(v1,v2)=(v2,v3);
+		}
+	}else{
+		// elseif wuDist == minDist3 then
+		if wu_dist*uv_len<uv_dist*wu_len{
+			(u,v)=(w,u);
+			uv=wu;
+			// uv_p=up_w; // unused
+			// b0, c0 = d0, b0
+			// b1, c1 = d1, b1
+			// before [a,b,c,d]
+			(v1,v2)=(v3,v1);
+			// after [a,d,b]
+		}
+	}
+
+	// local up = u:Cross(p)
+	// local pv = p:Cross(v)
+	// local uv_up = uv:Dot(up)
+	// local uv_pv = uv:Dot(pv)
+	let mut up=u.cross(p);
+	let pv=p.cross(v);
+	let uv_up=uv.dot(up);
+	let uv_pv=uv.dot(pv);
+
+	// if uv_up >= 0 and uv_pv >= 0 then
+	if !uv_up.is_negative()&&!uv_pv.is_negative(){
+		// local direction = uvw < 0 and uv or -uv
+		// return direction, a0, a1, b0, b1, c0, c1
+		let dir=if uv_w.is_negative(){
+			narrow_dir2(uv)
+		}else{
+			narrow_dir2(-uv)
+		};
+		return Reduce::Reduced(Reduced{
+			dir,
+			simplex:Simplex1_3::Simplex3([v0,v1,v2]),
+		});
+	}
+
+	// local u_u = u:Dot(u)
+	// local v_v = v:Dot(v)
+	// local uDist = uv_up/(u_u*v.magnitude)
+	// local vDist = uv_pv/(v_v*u.magnitude)
+	// local minDist2 = math.min(uDist, vDist)
+	let u_dist=uv_up*v.length();
+	let v_dist=uv_pv*u.length();
+
+	// if vDist == minDist2 then
+	if v_dist<u_dist{
+		u=v;
+		up=-pv;
+		uv=-uv;
+		// b0 = c0
+		// b1 = c1
+		v1=v2;
+	}
+
+	// local p_u = p:Dot(u)
+	let p_u=p.dot(u);
+
+	// if p_u >= 0 then
+	if !p_u.is_negative(){
+		// local direction = up:Cross(u)
+		let direction=up.cross(u);
+		// if direction.magnitude == 0 then
+		if direction==vec3::zero(){
+			// direction = uvw < 0 and uv or -uv
+			// return direction, a0, a1, b0, b1
+			let dir=if uv_w.is_negative(){
+				narrow_dir2(uv)
+			}else{
+				narrow_dir2(-uv)
+			};
+			return Reduce::Reduced(Reduced{
+				dir,
+				simplex:Simplex1_3::Simplex2([v0,v1]),
+			});
+		}
+
+		// return direction, a0, a1, b0, b1
+		return Reduce::Reduced(Reduced{
+			dir:narrow_dir3(direction),
+			simplex:Simplex1_3::Simplex2([v0,v1]),
+		});
+	}
+
+	// local direction = p
+	let dir=p;
+	// if direction.magnitude == 0 then
+	if dir==vec3::zero(){
+		// direction = uvw < 0 and uv or -uv
+		let dir=if uv_w.is_negative(){
+			narrow_dir2(uv)
+		}else{
+			narrow_dir2(-uv)
+		};
+		return Reduce::Reduced(Reduced{
+			dir,
+			simplex:Simplex1_3::Simplex1([v0]),
+		});
+	}
+
+	// return direction, a0, a1
+	Reduce::Reduced(Reduced{
+		dir,
+		simplex:Simplex1_3::Simplex1([v0]),
+	})
+}
+
+struct Reduced<Vert>{
+	dir:Planar64Vec3,
+	simplex:Simplex1_3<Vert>,
+}
+
+enum Reduce<Vert>{
+	Escape(Simplex<4,Vert>),
+	Reduced(Reduced<Vert>),
+}
+
+impl<Vert> Simplex2_4<Vert>{
+	fn reduce<M:MeshQuery<Vert=Vert,Position=Planar64Vec3>>(self,mesh:&M,point:Planar64Vec3)->Reduce<Vert>
+		where
+			M::Vert:Copy
+	{
+		match self{
+			Self::Simplex2(simplex)=>Reduce::Reduced(reduce2(simplex,mesh,point)),
+			Self::Simplex3(simplex)=>Reduce::Reduced(reduce3(simplex,mesh,point)),
+			Self::Simplex4(simplex)=>reduce4(simplex,mesh,point),
+		}
+	}
+}
+
+//infinity fev algorithm state transition
+#[derive(Debug)]
+enum Transition<Vert>{
+	Done,//found closest vert, no edges are better
+	Vert(Vert),//transition to vert
+}
+enum EV<M:MeshTopology>{
+	Vert(M::Vert),
+	Edge(M::Edge),
+}
+impl<M:MeshTopology> From<EV<M>> for FEV<M>{
+	fn from(value:EV<M>)->Self{
+		match value{
+			EV::Vert(minkowski_vert)=>FEV::Vert(minkowski_vert),
+			EV::Edge(minkowski_edge)=>FEV::Edge(minkowski_edge),
+		}
+	}
+}
+
+trait Contains{
+	fn contains(&self,point:Planar64Vec3)->bool;
+}
+
+// convenience type to check if a point is within some threshold of a plane.
+struct ThickPlane{
+	point:Planar64Vec3,
+	normal:Vector3<F128_64>,
+	epsilon:F192_96,
+}
+impl ThickPlane{
+	fn new<M:MeshQuery<Position=Planar64Vec3>>(mesh:&M,[v0,v1,v2]:Simplex<3,M::Vert>)->Self{
+		let p0=mesh.vert(v0);
+		let p1=mesh.vert(v1);
+		let p2=mesh.vert(v2);
+		let point=p0;
+		let normal=(p1-p0).cross(p2-p0);
+		// Allow ~ 2*sqrt(3) units of thickness on the plane
+		// This is to account for the variance of two voxels across the longest diagonal
+		let epsilon=(normal.length()*(Planar64::EPSILON*3)).wrap_192();
+		Self{point,normal,epsilon}
+	}
+}
+impl Contains for ThickPlane{
+	fn contains(&self,point:Planar64Vec3)->bool{
+		(point-self.point).dot(self.normal).abs()<=self.epsilon
+	}
+}
+
+struct ThickLine{
+	point:Planar64Vec3,
+	dir:Planar64Vec3,
+	epsilon:F256_128,
+}
+impl ThickLine{
+	fn new<M:MeshQuery<Position=Planar64Vec3>>(mesh:&M,[v0,v1]:Simplex<2,M::Vert>)->Self{
+		let p0=mesh.vert(v0);
+		let p1=mesh.vert(v1);
+		let point=p0;
+		let dir=p1-p0;
+		// Allow ~ 2*sqrt(3) units of thickness on the plane
+		// This is to account for the variance of two voxels across the longest diagonal
+		let epsilon=(dir.length_squared()*(Planar64::EPSILON*3)).widen_256();
+		Self{point,dir,epsilon}
+	}
+}
+impl Contains for ThickLine{
+	fn contains(&self,point:Planar64Vec3)->bool{
+		(point-self.point).cross(self.dir).length_squared()<=self.epsilon
+	}
+}
+
+struct EVFinder<'a,M,C>{
+	mesh:&'a M,
+	constraint:C,
+	best_distance_squared:F128_64,
+}
+
+impl<M:MeshQuery<Position=Planar64Vec3>,C:Contains> EVFinder<'_,M,C>
+	where
+		M::Vert:Copy,
+		M::DirectedEdge:Copy,
+{
+	fn next_transition_vert(&mut self,vert_id:M::Vert,point:Planar64Vec3)->Transition<M::Vert>{
+		let mut best_transition=Transition::Done;
+		self.mesh.for_each_vert_edge(vert_id,|directed_edge_id|{
+			//test if this edge's opposite vertex closer
+			let edge_verts=self.mesh.edge_verts(directed_edge_id.as_undirected());
+			//select opposite vertex
+			let test_vert_id=edge_verts.as_ref()[directed_edge_id.parity() as usize];
+			let test_pos=self.mesh.vert(test_vert_id);
+			let diff=point-test_pos;
+			let distance_squared=diff.dot(diff);
+			// ensure test_vert_id is coplanar to simplex
+			if distance_squared<self.best_distance_squared&&self.constraint.contains(test_pos){
+				best_transition=Transition::Vert(test_vert_id);
+				self.best_distance_squared=distance_squared;
+			}
+		});
+		best_transition
+	}
+	fn final_ev(&mut self,vert_id:M::Vert,point:Planar64Vec3)->EV<M>{
+		let mut best_transition=EV::Vert(vert_id);
+		let vert_pos=self.mesh.vert(vert_id);
+		let diff=point-vert_pos;
+		self.mesh.for_each_vert_edge(vert_id,|directed_edge_id|{
+			//test if this edge is closer
+			let edge_verts=self.mesh.edge_verts(directed_edge_id.as_undirected());
+			let test_vert_id=edge_verts.as_ref()[directed_edge_id.parity() as usize];
+			let test_pos=self.mesh.vert(test_vert_id);
+			let edge_n=test_pos-vert_pos;
+			let d=edge_n.dot(diff);
+			//test the edge
+			let edge_nn=edge_n.dot(edge_n);
+			// ensure edge contains closest point and directed_edge_id is coplanar to simplex
+			if !d.is_negative()&&d<=edge_nn&&self.constraint.contains(test_pos){
+				let distance_squared={
+					let c=diff.cross(edge_n);
+					//wrap for speed
+					(c.dot(c)/edge_nn).divide().wrap_128()
+				};
+				if distance_squared<=self.best_distance_squared{
+					best_transition=EV::Edge(directed_edge_id.as_undirected());
+					self.best_distance_squared=distance_squared;
+				}
+			}
+		});
+		best_transition
+	}
+	fn crawl_boundaries(&mut self,mut vert_id:M::Vert,point:Planar64Vec3)->EV<M>
+		where
+			M::Vert:Copy
+	{
+		loop{
+			match self.next_transition_vert(vert_id,point){
+				Transition::Done=>return self.final_ev(vert_id,point),
+				Transition::Vert(new_vert_id)=>vert_id=new_vert_id,
+			}
+		}
+	}
+}
+/// This function hops along parallel vertices until it finds the EV which contains the closest point to `point`.
+fn crawl_to_closest_ev<M:MeshQuery<Position=Planar64Vec3>>(mesh:&M,simplex:Simplex<2,M::Vert>,point:Planar64Vec3)->EV<M>
+	where
+		M::Vert:Copy,
+		M::DirectedEdge:Copy,
+{
+	// naively start at the closest vertex
+	// the closest vertex is not necessarily the one with the fewest boundary hops
+	// but it doesn't matter, we will get there regardless.
+	let (vert_id,best_distance_squared)=simplex.into_iter().map(|vert_id|{
+		let diff=point-mesh.vert(vert_id);
+		(vert_id,diff.dot(diff))
+	}).min_by_key(|&(_,d)|d).unwrap();
+
+	let constraint=ThickLine::new(mesh,simplex);
+	let mut finder=EVFinder{constraint,mesh,best_distance_squared};
+	//start on any vertex
+	//cross uncrossable vertex-edge boundaries until you find the closest vertex or edge
+	//cross edge-face boundary if it's uncrossable
+	finder.crawl_boundaries(vert_id,point)
+}
+
+/// This function hops along connected vertices until it finds the FEV which contains the closest point to `point`.
+fn crawl_to_closest_fev<'a>(mesh:&MinkowskiMesh<'a>,simplex:Simplex<3,MinkowskiVert>,point:Planar64Vec3)->FEV::<MinkowskiMesh<'a>>{
+	// naively start at the closest vertex
+	// the closest vertex is not necessarily the one with the fewest boundary hops
+	// but it doesn't matter, we will get there regardless.
+	let (vert_id,best_distance_squared)=simplex.into_iter().map(|vert_id|{
+		let diff=point-mesh.vert(vert_id);
+		(vert_id,diff.dot(diff))
+	}).min_by_key(|&(_,d)|d).unwrap();
+
+	let constraint=ThickPlane::new(mesh,simplex);
+	let mut finder=EVFinder{constraint,mesh,best_distance_squared};
+	//start on any vertex
+	//cross uncrossable vertex-edge boundaries until you find the closest vertex or edge
+	//cross edge-face boundary if it's uncrossable
+	match finder.crawl_boundaries(vert_id,point){
+		//if a vert is returned, it is the closest point to the infinity point
+		EV::Vert(vert_id)=>FEV::Vert(vert_id),
+		EV::Edge(edge_id)=>{
+			//cross to face if we are on the wrong side
+			let edge_n=mesh.edge_n(edge_id);
+			// point is multiplied by two because vert_sum sums two vertices.
+			let delta_pos=point*2-{
+				let &[v0,v1]=mesh.edge_verts(edge_id).as_ref();
+				mesh.vert(v0)+mesh.vert(v1)
+			};
+			for (i,&face_id) in mesh.edge_faces(edge_id).as_ref().iter().enumerate(){
+				//test if this face is closer
+				let (face_n,d)=mesh.face_nd(face_id);
+				//if test point is behind face, the face is invalid
+				// TODO: find out why I thought of this backwards
+				if !(face_n.dot(point)-d).is_positive(){
+					continue;
+				}
+				//edge-face boundary nd, n facing out of the face towards the edge
+				let boundary_n=face_n.cross(edge_n)*(i as i64*2-1);
+				let boundary_d=boundary_n.dot(delta_pos);
+				//is test point behind edge, i.e. contained in the face
+				if !boundary_d.is_positive(){
+					//both faces cannot pass this condition, return early if one does.
+					return FEV::Face(face_id);
+				}
+			}
+			FEV::Edge(edge_id)
+		},
+	}
+}
+
+pub fn closest_fev_not_inside<'a>(mesh:&MinkowskiMesh<'a>,point:Planar64Vec3)->Option<FEV<MinkowskiMesh<'a>>>{
+	const ENABLE_FAST_FAIL:bool=false;
+	// TODO: remove mesh negation
+	minimum_difference::<ENABLE_FAST_FAIL,_,_>(&-mesh,point,
+		// on_exact
+		|is_intersecting,simplex|{
+			if is_intersecting{
+				return None;
+			}
+			// Convert simplex to FEV
+			// Vertices must be inverted since the mesh is inverted
+			Some(match simplex{
+				Simplex1_3::Simplex1([v0])=>FEV::Vert(-v0),
+				Simplex1_3::Simplex2([v0,v1])=>{
+					// invert
+					let (v0,v1)=(-v0,-v1);
+					crawl_to_closest_ev(mesh,[v0,v1],point).into()
+				},
+				Simplex1_3::Simplex3([v0,v1,v2])=>{
+					// invert
+					let (v0,v1,v2)=(-v0,-v1,-v2);
+					// Shimmy to the side until you find a face that contains the closest point
+					// it's ALWAYS representable as a face, but this algorithm may
+					// return E or V in edge cases but I don't think that will break the face crawler
+					crawl_to_closest_fev(mesh,[v0,v1,v2],point)
+				},
+			})
+		},
+		// on_escape
+		|_simplex|{
+			// intersection is guaranteed at this point
+			// local norm, dist, u0, u1, v0, v1, w0, w1 = expand(queryP, queryQ, a0, a1, b0, b1, c0, c1, d0, d1, 1e-5)
+			// let simplex=refine_to_exact(mesh,simplex);
+			None
+		},
+		// fast_fail value is irrelevant and will never be returned!
+		||unreachable!()
+	)
+}
+
+pub fn contains_point(mesh:&MinkowskiMesh<'_>,point:Planar64Vec3)->bool{
+	const ENABLE_FAST_FAIL:bool=true;
+	// TODO: remove mesh negation
+	minimum_difference::<ENABLE_FAST_FAIL,_,_>(&-mesh,point,
+		// on_exact
+		|is_intersecting,_simplex|{
+			is_intersecting
+		},
+		// on_escape
+		|_simplex|{
+			// intersection is guaranteed at this point
+			true
+		},
+		// fast_fail value
+		||false
+	)
+}
+
+// local function minimumDifference(
+// 	queryP, radiusP,
+// 	queryQ, radiusQ,
+// 	exitRadius, testIntersection
+// )
+fn minimum_difference<const ENABLE_FAST_FAIL:bool,T,M:MeshQuery<Position=Planar64Vec3,Direction=Planar64Vec3>>(
+	mesh:&M,
+	point:Planar64Vec3,
+	on_exact:impl FnOnce(bool,Simplex1_3<M::Vert>)->T,
+	on_escape:impl FnOnce(Simplex<4,M::Vert>)->T,
+	on_fast_fail:impl FnOnce()->T,
+)->T
+	where
+		M::Vert:Copy
+{
+	// local initialAxis = queryQ() - queryP()
+	// local new_point_p = queryP(initialAxis)
+	// local new_point_q = queryQ(-initialAxis)
+	// local direction, a0, a1, b0, b1, c0, c1, d0, d1
+	let mut initial_axis=mesh.hint_point()+point;
+	// degenerate case
+	if initial_axis==vec3::zero(){
+		initial_axis=choose_any_direction();
+	}
+	let last_point=mesh.farthest_vert(-initial_axis);
+	// this represents the 'a' value in the commented code
+	let mut last_pos=mesh.vert(last_point);
+	let Reduced{dir:mut direction,simplex:mut simplex_small}=reduce1([last_point],mesh,point);
+
+	// exitRadius = testIntersection and 0 or exitRadius or 1/0
+	// for _ = 1, 100 do
+	loop{
+		// new_point_p = queryP(-direction)
+		// new_point_q = queryQ(direction)
+		// local next_point = new_point_q - new_point_p
+		let next_point=mesh.farthest_vert(direction);
+		let next_pos=mesh.vert(next_point);
+
+		// if -direction:Dot(next_point) > (exitRadius + radiusP + radiusQ)*direction.magnitude then
+		if ENABLE_FAST_FAIL&&direction.dot(next_pos+point).is_negative(){
+			return on_fast_fail();
+		}
+
+		let simplex_big=simplex_small.push_front(next_point);
+
+		// if
+		// 	direction:Dot(next_point - a) <= 0 or
+		// 	absDet(next_point, a, b, c) < 1e-6
+		if !direction.dot(next_pos-last_pos).is_positive()
+		||simplex_big.det_is_zero(mesh){
+			// Found enough information to compute the exact closest point.
+			// local norm = direction.unit
+			// local dist = a:Dot(norm)
+			// local hits = -dist < radiusP + radiusQ
+			let is_intersecting=(last_pos+point).dot(direction).is_positive();
+			return on_exact(is_intersecting,simplex_small);
+		}
+
+		// direction, a0, a1, b0, b1, c0, c1, d0, d1 = reduceSimplex(new_point_p, new_point_q, a0, a1, b0, b1, c0, c1)
+		match simplex_big.reduce(mesh,point){
+			// if a and b and c and d then
+			Reduce::Escape(simplex)=>{
+				// Enough information to conclude that the meshes are intersecting.
+				// Topology information is computed if needed.
+				return on_escape(simplex);
+			},
+			Reduce::Reduced(reduced)=>{
+				direction=reduced.dir;
+				simplex_small=reduced.simplex;
+			},
+		}
+
+		// next loop this will be a
+		last_pos=next_pos;
+	}
+}
+
+#[cfg(test)]
+mod test{
+	use super::*;
+    use crate::model::{PhysicsMesh,PhysicsMeshView};
+
+    fn mesh_contains_point(mesh:PhysicsMeshView<'_>,point:Planar64Vec3)->bool{
+		const ENABLE_FAST_FAIL:bool=true;
+		// TODO: remove mesh negation
+		minimum_difference::<ENABLE_FAST_FAIL,_,_>(&mesh,point,
+			// on_exact
+			|is_intersecting,_simplex|{
+				is_intersecting
+			},
+			// on_escape
+			|_simplex|{
+				// intersection is guaranteed at this point
+				true
+			},
+			// fast_fail value
+			||false
+		)
+	}
+
+	#[test]
+	fn test_cube_points(){
+		let mesh=PhysicsMesh::unit_cube();
+		let mesh_view=mesh.complete_mesh_view();
+		for x in -2..=2{
+			for y in -2..=2{
+				for z in -2..=2{
+					let point=vec3::int(x,y,z)>>1;
+					assert!(mesh_contains_point(mesh_view,point),"Mesh did not contain point {point}");
+				}
+			}
+		}
+	}
+}

engine/physics/src/minkowski.rs

@@ -0,0 +1,409 @@
+use core::ops::{Bound,RangeBounds};
+
+use strafesnet_common::integer::fixed_types::{F192_96,F256_128,F512_256};
+use strafesnet_common::integer::vec3::Vector3;
+use strafesnet_common::integer::{Planar64Vec3,Ratio,Fixed};
+use crate::model::into_giga_time;
+use crate::model::{SubmeshVertId,SubmeshEdgeId,SubmeshDirectedEdgeId,SubmeshFaceId,TransformedMesh,GigaTime};
+use crate::mesh_query::{MeshQuery,MeshTopology,DirectedEdge,UndirectedEdge};
+use crate::physics::{Time,Trajectory};
+
+struct AsRefHelper<T>(T);
+impl<T> AsRef<T> for AsRefHelper<T>{
+	fn as_ref(&self)->&T{
+		&self.0
+	}
+}
+
+//Note that a face on a minkowski mesh refers to a pair of fevs on the meshes it's summed from
+//(face,vertex)
+//(edge,edge)
+//(vertex,face)
+#[derive(Clone,Copy,Debug,Eq,PartialEq)]
+pub enum MinkowskiVert{
+	VertVert(SubmeshVertId,SubmeshVertId),
+}
+// TODO: remove this
+impl core::ops::Neg for MinkowskiVert{
+	type Output=Self;
+	fn neg(self)->Self::Output{
+		match self{
+			MinkowskiVert::VertVert(v0,v1)=>MinkowskiVert::VertVert(v1,v0),
+		}
+	}
+}
+#[derive(Clone,Copy,Debug)]
+pub enum MinkowskiEdge{
+	VertEdge(SubmeshVertId,SubmeshEdgeId),
+	EdgeVert(SubmeshEdgeId,SubmeshVertId),
+	//EdgeEdge when edges are parallel
+}
+impl UndirectedEdge for MinkowskiEdge{
+	type DirectedEdge=MinkowskiDirectedEdge;
+	fn as_directed(self,parity:bool)->Self::DirectedEdge{
+		match self{
+			MinkowskiEdge::VertEdge(v0,e1)=>MinkowskiDirectedEdge::VertEdge(v0,e1.as_directed(parity)),
+			MinkowskiEdge::EdgeVert(e0,v1)=>MinkowskiDirectedEdge::EdgeVert(e0.as_directed(parity),v1),
+		}
+	}
+}
+#[derive(Clone,Copy,Debug)]
+pub enum MinkowskiDirectedEdge{
+	VertEdge(SubmeshVertId,SubmeshDirectedEdgeId),
+	EdgeVert(SubmeshDirectedEdgeId,SubmeshVertId),
+	//EdgeEdge when edges are parallel
+}
+impl DirectedEdge for MinkowskiDirectedEdge{
+	type UndirectedEdge=MinkowskiEdge;
+	fn as_undirected(self)->Self::UndirectedEdge{
+		match self{
+			MinkowskiDirectedEdge::VertEdge(v0,e1)=>MinkowskiEdge::VertEdge(v0,e1.as_undirected()),
+			MinkowskiDirectedEdge::EdgeVert(e0,v1)=>MinkowskiEdge::EdgeVert(e0.as_undirected(),v1),
+		}
+	}
+	fn parity(&self)->bool{
+		match self{
+			MinkowskiDirectedEdge::VertEdge(_,e)
+			|MinkowskiDirectedEdge::EdgeVert(e,_)=>e.parity(),
+		}
+	}
+}
+#[derive(Clone,Copy,Debug,Hash)]
+pub enum MinkowskiFace{
+	VertFace(SubmeshVertId,SubmeshFaceId),
+	EdgeEdge(SubmeshEdgeId,SubmeshEdgeId,bool),
+	FaceVert(SubmeshFaceId,SubmeshVertId),
+	//EdgeFace
+	//FaceEdge
+	//FaceFace
+}
+
+#[derive(Debug)]
+pub struct MinkowskiMesh<'a>{
+	mesh0:TransformedMesh<'a>,
+	mesh1:TransformedMesh<'a>,
+}
+
+// TODO: remove this
+impl<'a> core::ops::Neg for &MinkowskiMesh<'a>{
+	type Output=MinkowskiMesh<'a>;
+	fn neg(self)->Self::Output{
+		MinkowskiMesh::minkowski_sum(self.mesh1,self.mesh0)
+	}
+}
+
+impl MinkowskiMesh<'_>{
+	pub fn minkowski_sum<'a>(mesh0:TransformedMesh<'a>,mesh1:TransformedMesh<'a>)->MinkowskiMesh<'a>{
+		MinkowskiMesh{
+			mesh0,
+			mesh1,
+		}
+	}
+	pub fn predict_collision_in(&self,trajectory:&Trajectory,range:impl RangeBounds<Time>)->Option<(MinkowskiFace,GigaTime)>{
+		let start_position=match range.start_bound(){
+			Bound::Included(time)=>trajectory.extrapolated_position(*time),
+			Bound::Excluded(time)=>trajectory.extrapolated_position(*time),
+			Bound::Unbounded=>trajectory.position,
+		};
+		let fev=crate::minimum_difference::closest_fev_not_inside(self,start_position)?;
+		//continue forwards along the body parabola
+		fev.crawl(self,trajectory,range.start_bound(),range.end_bound()).hit()
+	}
+	pub fn predict_collision_out(&self,trajectory:&Trajectory,range:impl RangeBounds<Time>)->Option<(MinkowskiFace,GigaTime)>{
+		let (lower_bound,upper_bound)=(range.start_bound(),range.end_bound());
+		// TODO: handle unbounded collision using infinity fev
+		let start_position=match upper_bound{
+			Bound::Included(time)=>trajectory.extrapolated_position(*time),
+			Bound::Excluded(time)=>trajectory.extrapolated_position(*time),
+			Bound::Unbounded=>trajectory.position,
+		};
+		let fev=crate::minimum_difference::closest_fev_not_inside(self,start_position)?;
+		// swap and negate bounds to do a time inversion
+		let (lower_bound,upper_bound)=(upper_bound.map(|&t|-t),lower_bound.map(|&t|-t));
+		let time_reversed_trajectory=-trajectory;
+		//continue backwards along the body parabola
+		fev.crawl(self,&time_reversed_trajectory,lower_bound.as_ref(),upper_bound.as_ref()).hit()
+			//no need to test -time<time_limit because of the first step
+			.map(|(face,time)|(face,-time))
+	}
+	pub fn predict_collision_face_out(&self,trajectory:&Trajectory,range:impl RangeBounds<Time>,contact_face_id:MinkowskiFace)->Option<(MinkowskiDirectedEdge,GigaTime)>{
+		// TODO: make better
+		use crate::face_crawler::{low,upp};
+		//no algorithm needed, there is only one state and two cases (Edge,None)
+		//determine when it passes an edge ("sliding off" case)
+		let start_time=range.start_bound().map(|&t|(t-trajectory.time).to_ratio());
+		let mut best_time=range.end_bound().map(|&t|into_giga_time(t,trajectory.time));
+		let mut best_edge=None;
+		let face_n=self.face_nd(contact_face_id).0;
+		self.for_each_face_edge(contact_face_id,|directed_edge_id|{
+			let edge_n=self.directed_edge_n(directed_edge_id);
+			//f x e points in
+			let n=face_n.cross(edge_n);
+			let &[v0,v1]=self.edge_verts(directed_edge_id.as_undirected()).as_ref();
+			let d=n.dot(self.vert(v0)+self.vert(v1));
+			//WARNING! d outside of *2
+			//WARNING: truncated precision
+			//wrap for speed
+			for dt in F256_128::zeroes2(((n.dot(trajectory.position))*2-d).wrap_256(),n.dot(trajectory.velocity).wrap_256()*2,n.dot(trajectory.acceleration).wrap_256()){
+				if low(&start_time,&dt)&&upp(&dt,&best_time)&&n.dot(trajectory.extrapolated_velocity_ratio_dt(dt)).is_negative(){
+					best_time=Bound::Included(dt);
+					best_edge=Some((directed_edge_id,dt));
+					break;
+				}
+			}
+		});
+		best_edge
+	}
+	pub fn contains_point(&self,point:Planar64Vec3)->bool{
+		crate::minimum_difference::contains_point(self,point)
+	}
+}
+impl MeshQuery for MinkowskiMesh<'_>{
+	type Direction=Planar64Vec3;
+	type Position=Planar64Vec3;
+	type Normal=Vector3<F192_96>;
+	type Offset=F256_128;
+	// TODO: relative d
+	fn face_nd(&self,face_id:MinkowskiFace)->(Self::Normal,Self::Offset){
+		match face_id{
+			MinkowskiFace::VertFace(v0,f1)=>{
+				let (n,d)=self.mesh1.face_nd(f1);
+				(-n,d-n.dot(self.mesh0.vert(v0)))
+			},
+			MinkowskiFace::EdgeEdge(e0,e1,parity)=>{
+				let edge0_n=self.mesh0.edge_n(e0);
+				let edge1_n=self.mesh1.edge_n(e1);
+				let &[e0v0,e0v1]=self.mesh0.edge_verts(e0).as_ref();
+				let &[e1v0,e1v1]=self.mesh1.edge_verts(e1).as_ref();
+				let n=edge0_n.cross(edge1_n);
+				let e0d=n.dot(self.mesh0.vert(e0v0)+self.mesh0.vert(e0v1));
+				let e1d=n.dot(self.mesh1.vert(e1v0)+self.mesh1.vert(e1v1));
+				((n*(parity as i64*4-2)).widen_192(),((e0d-e1d)*(parity as i64*2-1)).widen_256())
+			},
+			MinkowskiFace::FaceVert(f0,v1)=>{
+				let (n,d)=self.mesh0.face_nd(f0);
+				(n,d-n.dot(self.mesh1.vert(v1)))
+			},
+		}
+	}
+	fn vert(&self,vert_id:MinkowskiVert)->Planar64Vec3{
+		match vert_id{
+			MinkowskiVert::VertVert(v0,v1)=>{
+				self.mesh0.vert(v0)-self.mesh1.vert(v1)
+			},
+		}
+	}
+	fn hint_point(&self)->Planar64Vec3{
+		self.mesh0.hint_point()-self.mesh1.hint_point()
+	}
+	fn farthest_vert(&self,dir:Planar64Vec3)->MinkowskiVert{
+		MinkowskiVert::VertVert(self.mesh0.farthest_vert(dir),self.mesh1.farthest_vert(-dir))
+	}
+	fn edge_n(&self,edge_id:Self::Edge)->Self::Direction{
+		let &[v0,v1]=self.edge_verts(edge_id).as_ref();
+		self.vert(v1)-self.vert(v0)
+	}
+	fn directed_edge_n(&self,directed_edge_id:Self::DirectedEdge)->Self::Direction{
+		let &[v0,v1]=self.edge_verts(directed_edge_id.as_undirected()).as_ref();
+		(self.vert(v1)-self.vert(v0))*((directed_edge_id.parity() as i64)*2-1)
+	}
+}
+impl MeshTopology for MinkowskiMesh<'_>{
+	type Face=MinkowskiFace;
+	type Edge=MinkowskiEdge;
+	type DirectedEdge=MinkowskiDirectedEdge;
+	type Vert=MinkowskiVert;
+	fn for_each_vert_edge(&self,vert_id:Self::Vert,mut f:impl FnMut(Self::DirectedEdge)){
+		match vert_id{
+			MinkowskiVert::VertVert(v0,v1)=>{
+				//detect shared volume when the other mesh is mirrored along a test edge dir
+				let v0f={
+					let mut faces=Vec::new();
+					self.mesh0.for_each_vert_face(v0,|face|faces.push(face));
+					faces
+				};
+				let v1f={
+					let mut faces=Vec::new();
+					self.mesh1.for_each_vert_face(v1,|face|faces.push(face));
+					faces
+				};
+				let v0f_n:Vec<_>=v0f.iter().map(|&face_id|self.mesh0.face_nd(face_id).0).collect();
+				let v1f_n:Vec<_>=v1f.iter().map(|&face_id|self.mesh1.face_nd(face_id).0).collect();
+				// scratch vector
+				let mut face_normals=Vec::with_capacity(v0f.len()+v1f.len());
+				face_normals.clone_from(&v0f_n);
+				self.mesh0.for_each_vert_edge(v0,|directed_edge_id|{
+					let n=self.mesh0.directed_edge_n(directed_edge_id);
+					let nn=n.dot(n);
+					// TODO: there's gotta be a better way to do this
+					// drop faces beyond v0f_n
+					face_normals.truncate(v0f.len());
+					// make a set of faces from mesh0's perspective
+					for face_n in &v1f_n{
+						//add reflected mesh1 faces
+						//wrap for speed
+						face_normals.push(*face_n-(n*face_n.dot(n)*2/nn).divide().wrap_192());
+					}
+					if is_empty_volume(&face_normals){
+						f(MinkowskiDirectedEdge::EdgeVert(directed_edge_id,v1));
+					}
+				});
+				face_normals.clone_from(&v1f_n);
+				self.mesh1.for_each_vert_edge(v1,|directed_edge_id|{
+					let n=self.mesh1.directed_edge_n(directed_edge_id);
+					let nn=n.dot(n);
+					// drop faces beyond v1f_n
+					face_normals.truncate(v1f.len());
+					// make a set of faces from mesh1's perspective
+					for face_n in &v0f_n{
+						//wrap for speed
+						face_normals.push(*face_n-(n*face_n.dot(n)*2/nn).divide().wrap_192());
+					}
+					if is_empty_volume(&face_normals){
+						f(MinkowskiDirectedEdge::VertEdge(v0,directed_edge_id));
+					}
+				});
+			},
+		}
+	}
+	fn for_each_vert_face(&self,_vert_id:Self::Vert,_f:impl FnMut(Self::Face)){
+		unimplemented!()
+	}
+	fn edge_faces(&self,edge_id:Self::Edge)->impl AsRef<[Self::Face;2]>{
+		match edge_id{
+			MinkowskiEdge::VertEdge(v0,e1)=>{
+				//faces are listed backwards from the minkowski mesh
+				let v0e={
+					let mut edges=Vec::new();
+					self.mesh0.for_each_vert_edge(v0,|edge|edges.push(edge));
+					edges
+				};
+				let &[e1f0,e1f1]=self.mesh1.edge_faces(e1).as_ref();
+				AsRefHelper([(e1f1,false),(e1f0,true)].map(|(edge_face_id1,face_parity)|{
+					let mut best_edge=None;
+					let mut best_d:Ratio<F512_256,F512_256>=Ratio::new(Fixed::ZERO,Fixed::ONE);
+					let edge_face1_n=self.mesh1.face_nd(edge_face_id1).0;
+					let edge_face1_nn=edge_face1_n.dot(edge_face1_n);
+					for &directed_edge_id0 in &v0e{
+						let edge0_n=self.mesh0.directed_edge_n(directed_edge_id0);
+						//must be behind other face.
+						let d=edge_face1_n.dot(edge0_n);
+						if d.is_negative(){
+							let edge0_nn=edge0_n.dot(edge0_n);
+							// Assume not every number is huge
+							// TODO: revisit this
+							let dd=(d*d)/(edge_face1_nn*edge0_nn);
+							if best_d<dd{
+								best_d=dd;
+								best_edge=Some(directed_edge_id0);
+							}
+						}
+					}
+					best_edge.map_or(
+						MinkowskiFace::VertFace(v0,edge_face_id1),
+						|directed_edge_id0|MinkowskiFace::EdgeEdge(directed_edge_id0.as_undirected(),e1,directed_edge_id0.parity()^face_parity)
+					)
+				}))
+			},
+			MinkowskiEdge::EdgeVert(e0,v1)=>{
+				//tracking index with an external variable because .enumerate() is not available
+				let v1e={
+					let mut edges=Vec::new();
+					self.mesh1.for_each_vert_edge(v1,|edge|edges.push(edge));
+					edges
+				};
+				let &[e0f0,e0f1]=self.mesh0.edge_faces(e0).as_ref();
+				AsRefHelper([(e0f0,true),(e0f1,false)].map(|(edge_face_id0,face_parity)|{
+					let mut best_edge=None;
+					let mut best_d:Ratio<F512_256,F512_256>=Ratio::new(Fixed::ZERO,Fixed::ONE);
+					let edge_face0_n=self.mesh0.face_nd(edge_face_id0).0;
+					let edge_face0_nn=edge_face0_n.dot(edge_face0_n);
+					for &directed_edge_id1 in &v1e{
+						let edge1_n=self.mesh1.directed_edge_n(directed_edge_id1);
+						let d=edge_face0_n.dot(edge1_n);
+						if d.is_negative(){
+							let edge1_nn=edge1_n.dot(edge1_n);
+							let dd=(d*d)/(edge_face0_nn*edge1_nn);
+							if best_d<dd{
+								best_d=dd;
+								best_edge=Some(directed_edge_id1);
+							}
+						}
+					}
+					best_edge.map_or(
+						MinkowskiFace::FaceVert(edge_face_id0,v1),
+						|directed_edge_id1|MinkowskiFace::EdgeEdge(e0,directed_edge_id1.as_undirected(),directed_edge_id1.parity()^face_parity)
+					)
+				}))
+			},
+		}
+	}
+	fn edge_verts(&self,edge_id:Self::Edge)->impl AsRef<[Self::Vert;2]>{
+		AsRefHelper(match edge_id{
+			MinkowskiEdge::VertEdge(v0,e1)=>self.mesh1.edge_verts(e1).as_ref().map(|vert_id1|
+				MinkowskiVert::VertVert(v0,vert_id1)
+			),
+			MinkowskiEdge::EdgeVert(e0,v1)=>self.mesh0.edge_verts(e0).as_ref().map(|vert_id0|
+				MinkowskiVert::VertVert(vert_id0,v1)
+			),
+		})
+	}
+	fn for_each_face_vert(&self,_face_id:Self::Face,_f:impl FnMut(Self::Vert)){
+		unimplemented!()
+	}
+	fn for_each_face_edge(&self,face_id:Self::Face,mut f:impl FnMut(Self::DirectedEdge)){
+		match face_id{
+			MinkowskiFace::VertFace(v0,f1)=>{
+				self.mesh1.for_each_face_edge(f1,|edge_id1|
+					f(MinkowskiDirectedEdge::VertEdge(v0,edge_id1.reverse()))
+				)
+			},
+			MinkowskiFace::EdgeEdge(e0,e1,parity)=>{
+				let &[e0v0,e0v1]=self.mesh0.edge_verts(e0).as_ref();
+				let &[e1v0,e1v1]=self.mesh1.edge_verts(e1).as_ref();
+				//could sort this if ordered edges are needed
+				//probably just need to reverse this list according to parity
+				f(MinkowskiDirectedEdge::VertEdge(e0v0,e1.as_directed(parity)));
+				f(MinkowskiDirectedEdge::EdgeVert(e0.as_directed(!parity),e1v0));
+				f(MinkowskiDirectedEdge::VertEdge(e0v1,e1.as_directed(!parity)));
+				f(MinkowskiDirectedEdge::EdgeVert(e0.as_directed(parity),e1v1));
+			},
+			MinkowskiFace::FaceVert(f0,v1)=>{
+				self.mesh0.for_each_face_edge(f0,|edge_id0|
+					f(MinkowskiDirectedEdge::EdgeVert(edge_id0,v1))
+				)
+			},
+		}
+	}
+}
+
+fn is_empty_volume(normals:&[Vector3<F192_96>])->bool{
+	let len=normals.len();
+	for i in 0..len-1{
+		for j in i+1..len{
+			let n=normals[i].cross(normals[j]);
+			let mut d_comp=None;
+			for k in 0..len{
+				if k!=i&&k!=j{
+					let d=n.dot(normals[k]).is_negative();
+					if let &Some(comp)=&d_comp{
+						// This is testing if d_comp*d < 0
+						if comp^d{
+							return true;
+						}
+					}else{
+						d_comp=Some(d);
+					}
+				}
+			}
+		}
+	}
+	return false;
+}
+
+#[test]
+fn test_is_empty_volume(){
+	use strafesnet_common::integer::vec3;
+	assert!(!is_empty_volume(&[vec3::X.widen_192(),vec3::Y.widen_192(),vec3::Z.widen_192()]));
+	assert!(is_empty_volume(&[vec3::X.widen_192(),vec3::Y.widen_192(),vec3::Z.widen_192(),vec3::NEG_X.widen_192()]));
+}

engine/physics/src/model.rs

@@ -0,0 +1,570 @@
+use std::collections::{HashSet,HashMap};
+use strafesnet_common::integer::fixed_types::{F128_64,F192_96,F256_128};
+use strafesnet_common::integer::vec3::Vector3;
+use strafesnet_common::model::{self,MeshId,PolygonIter};
+use strafesnet_common::integer::{self,vec3,Fixed,Planar64,Planar64Vec3,Ratio};
+use strafesnet_common::physics::Time;
+use crate::mesh_query::{MeshQuery,MeshTopology,DirectedEdge,UndirectedEdge};
+
+struct AsRefHelper<T>(T);
+impl<T> AsRef<T> for AsRefHelper<T>{
+	fn as_ref(&self)->&T{
+		&self.0
+	}
+}
+
+#[derive(Debug,Clone,Copy,Hash,id::Id,Eq,PartialEq)]
+pub struct MeshVertId(u32);
+#[derive(Debug,Clone,Copy,Hash,id::Id,Eq,PartialEq)]
+pub struct MeshFaceId(u32);
+
+#[derive(Debug,Clone,Copy,Hash,id::Id,Eq,PartialEq)]
+pub struct SubmeshVertId(u32);
+#[derive(Debug,Clone,Copy,Hash,id::Id,Eq,PartialEq)]
+pub struct SubmeshEdgeId(u32);
+/// DirectedEdgeId refers to an EdgeId when undirected.
+#[derive(Debug,Clone,Copy,Hash,id::Id,Eq,PartialEq)]
+pub struct SubmeshDirectedEdgeId(u32);
+#[derive(Debug,Clone,Copy,Hash,id::Id,Eq,PartialEq)]
+pub struct SubmeshFaceId(u32);
+
+impl UndirectedEdge for SubmeshEdgeId{
+	type DirectedEdge=SubmeshDirectedEdgeId;
+	fn as_directed(self,parity:bool)->SubmeshDirectedEdgeId{
+		SubmeshDirectedEdgeId(self.0|((parity as u32)<<(u32::BITS-1)))
+	}
+}
+impl DirectedEdge for SubmeshDirectedEdgeId{
+	type UndirectedEdge=SubmeshEdgeId;
+	fn as_undirected(self)->SubmeshEdgeId{
+		SubmeshEdgeId(self.0&!(1<<(u32::BITS-1)))
+	}
+	fn parity(&self)->bool{
+		self.0&(1<<(u32::BITS-1))!=0
+	}
+}
+
+//use Unit32 #[repr(C)] for map files
+#[derive(Clone,Copy,Debug,Hash,Eq,PartialEq)]
+struct Face{
+	normal:Planar64Vec3,
+	dot:Planar64,
+}
+#[derive(Debug)]
+struct Vert(Planar64Vec3);
+#[derive(Debug)]
+struct FaceRefs{
+	// I didn't write it down, but I assume the edges are directed
+	// clockwise when looking towards the face normal, i.e. right hand rule.
+	edges:Vec<SubmeshDirectedEdgeId>,
+	//verts are redundant, use edge[i].verts[0]
+	//verts:Vec<VertId>,
+}
+#[derive(Debug)]
+struct EdgeRefs{
+	faces:[SubmeshFaceId;2],//left, right
+	verts:[SubmeshVertId;2],//start, end
+}
+#[derive(Debug)]
+struct VertRefs{
+	faces:Vec<SubmeshFaceId>,
+	// edges are always directed away from the vert
+	edges:Vec<SubmeshDirectedEdgeId>,
+}
+#[derive(Debug)]
+pub struct PhysicsMeshData{
+	//this contains all real and virtual faces used in both the complete mesh and convex submeshes
+	//faces are sorted such that all faces that belong to the complete mesh appear first, and then
+	//all remaining faces are virtual to operate internal logic of the face crawler
+	//and cannot be part of a physics collision
+	//virtual faces are only used in convex submeshes.
+	faces:Vec<Face>,//MeshFaceId indexes this list
+	verts:Vec<Vert>,//MeshVertId indexes this list
+}
+#[derive(Debug)]
+pub struct PhysicsMeshTopology{
+	//mapping of local ids to PhysicsMeshData ids
+	faces:Vec<MeshFaceId>,//SubmeshFaceId indexes this list
+	verts:Vec<MeshVertId>,//SubmeshVertId indexes this list
+	//all ids here are local to this object
+	face_topology:Vec<FaceRefs>,
+	edge_topology:Vec<EdgeRefs>,
+	vert_topology:Vec<VertRefs>,
+}
+#[derive(Clone,Copy,Hash,id::Id,Eq,PartialEq)]
+pub struct PhysicsMeshId(u32);
+impl Into<MeshId> for PhysicsMeshId{
+	fn into(self)->MeshId{
+		MeshId::new(self.0)
+	}
+}
+impl From<MeshId> for PhysicsMeshId{
+	fn from(value:MeshId)->Self{
+		Self::new(value.get())
+	}
+}
+#[derive(Debug,Default,Clone,Copy,Hash,id::Id,Eq,PartialEq)]
+pub struct PhysicsSubmeshId(u32);
+pub struct PhysicsMesh{
+	data:PhysicsMeshData,
+	// The complete mesh is unused at this time.
+	// complete_mesh:PhysicsMeshTopology,
+	// Submeshes are guaranteed to be convex and may contain
+	// "virtual" faces which are not part of the complete mesh.
+	// Physics calculations should never resolve to hitting
+	// a virtual face.
+	submeshes:Vec<PhysicsMeshTopology>,
+}
+impl PhysicsMesh{
+	pub fn unit_cube()->Self{
+		//go go gadget debug print mesh
+		let data=PhysicsMeshData{
+			faces:vec![
+				Face{normal:vec3::raw_xyz( 4294967296, 0, 0),dot:Planar64::raw(4294967296)},
+				Face{normal:vec3::raw_xyz( 0, 4294967296, 0),dot:Planar64::raw(4294967296)},
+				Face{normal:vec3::raw_xyz( 0, 0, 4294967296),dot:Planar64::raw(4294967296)},
+				Face{normal:vec3::raw_xyz(-4294967296, 0, 0),dot:Planar64::raw(4294967296)},
+				Face{normal:vec3::raw_xyz( 0,-4294967296, 0),dot:Planar64::raw(4294967296)},
+				Face{normal:vec3::raw_xyz( 0, 0,-4294967296),dot:Planar64::raw(4294967296)}
+			],
+			verts:vec![
+				Vert(vec3::raw_xyz( 4294967296,-4294967296,-4294967296)),
+				Vert(vec3::raw_xyz( 4294967296, 4294967296,-4294967296)),
+				Vert(vec3::raw_xyz( 4294967296, 4294967296, 4294967296)),
+				Vert(vec3::raw_xyz( 4294967296,-4294967296, 4294967296)),
+				Vert(vec3::raw_xyz(-4294967296, 4294967296,-4294967296)),
+				Vert(vec3::raw_xyz(-4294967296, 4294967296, 4294967296)),
+				Vert(vec3::raw_xyz(-4294967296,-4294967296, 4294967296)),
+				Vert(vec3::raw_xyz(-4294967296,-4294967296,-4294967296))
+			]
+		};
+		let mesh_topology=PhysicsMeshTopology{
+			faces:(0..data.faces.len() as u32).map(MeshFaceId::new).collect(),
+			verts:(0..data.verts.len() as u32).map(MeshVertId::new).collect(),
+			face_topology:vec![
+				FaceRefs{edges:vec![SubmeshDirectedEdgeId((9223372036854775808u64-(1<<63)+(1<<31)) as u32),SubmeshDirectedEdgeId((9223372036854775809u64-(1<<63)+(1<<31)) as u32),SubmeshDirectedEdgeId((9223372036854775810u64-(1<<63)+(1<<31)) as u32),SubmeshDirectedEdgeId(3)]},
+				FaceRefs{edges:vec![SubmeshDirectedEdgeId((9223372036854775812u64-(1<<63)+(1<<31)) as u32),SubmeshDirectedEdgeId((9223372036854775813u64-(1<<63)+(1<<31)) as u32),SubmeshDirectedEdgeId(6),SubmeshDirectedEdgeId(1)]},
+				FaceRefs{edges:vec![SubmeshDirectedEdgeId(7),SubmeshDirectedEdgeId(2),SubmeshDirectedEdgeId((9223372036854775814u64-(1<<63)+(1<<31)) as u32),SubmeshDirectedEdgeId((9223372036854775816u64-(1<<63)+(1<<31)) as u32)]},
+				FaceRefs{edges:vec![SubmeshDirectedEdgeId(8),SubmeshDirectedEdgeId(5),SubmeshDirectedEdgeId((9223372036854775817u64-(1<<63)+(1<<31)) as u32),SubmeshDirectedEdgeId(10)]},
+				FaceRefs{edges:vec![SubmeshDirectedEdgeId((9223372036854775815u64-(1<<63)+(1<<31)) as u32),SubmeshDirectedEdgeId((9223372036854775818u64-(1<<63)+(1<<31)) as u32),SubmeshDirectedEdgeId(11),SubmeshDirectedEdgeId((9223372036854775811u64-(1<<63)+(1<<31)) as u32)]},
+				FaceRefs{edges:vec![SubmeshDirectedEdgeId(4),SubmeshDirectedEdgeId(0),SubmeshDirectedEdgeId((9223372036854775819u64-(1<<63)+(1<<31)) as u32),SubmeshDirectedEdgeId(9)]}
+			],
+			edge_topology:vec![
+				EdgeRefs{faces:[SubmeshFaceId(0),SubmeshFaceId(5)],verts:[SubmeshVertId(0),SubmeshVertId(1)]},
+				EdgeRefs{faces:[SubmeshFaceId(0),SubmeshFaceId(1)],verts:[SubmeshVertId(1),SubmeshVertId(2)]},
+				EdgeRefs{faces:[SubmeshFaceId(0),SubmeshFaceId(2)],verts:[SubmeshVertId(2),SubmeshVertId(3)]},
+				EdgeRefs{faces:[SubmeshFaceId(4),SubmeshFaceId(0)],verts:[SubmeshVertId(0),SubmeshVertId(3)]},
+				EdgeRefs{faces:[SubmeshFaceId(1),SubmeshFaceId(5)],verts:[SubmeshVertId(1),SubmeshVertId(4)]},
+				EdgeRefs{faces:[SubmeshFaceId(1),SubmeshFaceId(3)],verts:[SubmeshVertId(4),SubmeshVertId(5)]},
+				EdgeRefs{faces:[SubmeshFaceId(2),SubmeshFaceId(1)],verts:[SubmeshVertId(2),SubmeshVertId(5)]},
+				EdgeRefs{faces:[SubmeshFaceId(4),SubmeshFaceId(2)],verts:[SubmeshVertId(3),SubmeshVertId(6)]},
+				EdgeRefs{faces:[SubmeshFaceId(2),SubmeshFaceId(3)],verts:[SubmeshVertId(5),SubmeshVertId(6)]},
+				EdgeRefs{faces:[SubmeshFaceId(3),SubmeshFaceId(5)],verts:[SubmeshVertId(4),SubmeshVertId(7)]},
+				EdgeRefs{faces:[SubmeshFaceId(4),SubmeshFaceId(3)],verts:[SubmeshVertId(6),SubmeshVertId(7)]},
+				EdgeRefs{faces:[SubmeshFaceId(5),SubmeshFaceId(4)],verts:[SubmeshVertId(0),SubmeshVertId(7)]}
+			],
+			vert_topology:vec![
+				VertRefs{faces:vec![SubmeshFaceId(0),SubmeshFaceId(4),SubmeshFaceId(5)],edges:vec![SubmeshDirectedEdgeId((9223372036854775811u64-(1<<63)+(1<<31)) as u32),SubmeshDirectedEdgeId((9223372036854775819u64-(1<<63)+(1<<31)) as u32),SubmeshDirectedEdgeId((9223372036854775808u64-(1<<63)+(1<<31)) as u32)]},
+				VertRefs{faces:vec![SubmeshFaceId(0),SubmeshFaceId(5),SubmeshFaceId(1)],edges:vec![SubmeshDirectedEdgeId((9223372036854775812u64-(1<<63)+(1<<31)) as u32),SubmeshDirectedEdgeId(0),SubmeshDirectedEdgeId((9223372036854775809u64-(1<<63)+(1<<31)) as u32)]},
+				VertRefs{faces:vec![SubmeshFaceId(0),SubmeshFaceId(2),SubmeshFaceId(1)],edges:vec![SubmeshDirectedEdgeId(1),SubmeshDirectedEdgeId((9223372036854775810u64-(1<<63)+(1<<31)) as u32),SubmeshDirectedEdgeId((9223372036854775814u64-(1<<63)+(1<<31)) as u32)]},
+				VertRefs{faces:vec![SubmeshFaceId(0),SubmeshFaceId(2),SubmeshFaceId(4)],edges:vec![SubmeshDirectedEdgeId(2),SubmeshDirectedEdgeId(3),SubmeshDirectedEdgeId((9223372036854775815u64-(1<<63)+(1<<31)) as u32)]},
+				VertRefs{faces:vec![SubmeshFaceId(3),SubmeshFaceId(5),SubmeshFaceId(1)],edges:vec![SubmeshDirectedEdgeId(4),SubmeshDirectedEdgeId((9223372036854775817u64-(1<<63)+(1<<31)) as u32),SubmeshDirectedEdgeId((9223372036854775813u64-(1<<63)+(1<<31)) as u32)]},
+				VertRefs{faces:vec![SubmeshFaceId(2),SubmeshFaceId(3),SubmeshFaceId(1)],edges:vec![SubmeshDirectedEdgeId(5),SubmeshDirectedEdgeId(6),SubmeshDirectedEdgeId((9223372036854775816u64-(1<<63)+(1<<31)) as u32)]},
+				VertRefs{faces:vec![SubmeshFaceId(2),SubmeshFaceId(3),SubmeshFaceId(4)],edges:vec![SubmeshDirectedEdgeId(7),SubmeshDirectedEdgeId(8),SubmeshDirectedEdgeId((9223372036854775818u64-(1<<63)+(1<<31)) as u32)]},
+				VertRefs{faces:vec![SubmeshFaceId(4),SubmeshFaceId(3),SubmeshFaceId(5)],edges:vec![SubmeshDirectedEdgeId(10),SubmeshDirectedEdgeId(11),SubmeshDirectedEdgeId(9)]}
+			]
+		};
+		Self{
+			data,
+			// complete_mesh:mesh_topology.clone(),
+			submeshes:vec![mesh_topology],
+		}
+	}
+	pub fn unit_cylinder()->Self{
+		Self::unit_cube()
+	}
+	#[inline]
+	pub fn complete_mesh(&self)->&PhysicsMeshTopology{
+		// If there is exactly one submesh, then the complete mesh is identical to it.
+		if self.submeshes.len()==1{
+			self.submeshes.first().unwrap()
+		}else{
+			panic!("PhysicsMesh complete mesh is not known");
+		}
+	}
+	#[inline]
+	pub fn complete_mesh_view(&self)->PhysicsMeshView<'_>{
+		PhysicsMeshView{
+			data:&self.data,
+			topology:self.complete_mesh(),
+		}
+	}
+	#[inline]
+	pub fn submeshes(&self)->&[PhysicsMeshTopology]{
+		&self.submeshes
+	}
+	#[inline]
+	pub fn submesh_view(&self,submesh_id:PhysicsSubmeshId)->PhysicsMeshView<'_>{
+		PhysicsMeshView{
+			data:&self.data,
+			topology:&self.submeshes()[submesh_id.get() as usize],
+		}
+	}
+	pub fn submesh_views(&self)->impl Iterator<Item=PhysicsMeshView<'_>>{
+		self.submeshes().iter().map(|topology|PhysicsMeshView{
+			data:&self.data,
+			topology,
+		})
+	}
+}
+
+//mesh builder code
+#[derive(Default,Clone)]
+struct VertRefGuy{
+	edges:HashSet<SubmeshDirectedEdgeId>,
+	faces:HashSet<SubmeshFaceId>,
+}
+#[derive(Clone,Hash,Eq,PartialEq)]
+struct EdgeRefVerts([SubmeshVertId;2]);
+impl EdgeRefVerts{
+	const fn new(v0:SubmeshVertId,v1:SubmeshVertId)->(Self,bool){
+		(if v0.0<v1.0{
+			Self([v0,v1])
+		}else{
+			Self([v1,v0])
+		},v0.0<v1.0)
+	}
+}
+struct EdgeRefFaces([SubmeshFaceId;2]);
+impl EdgeRefFaces{
+	const fn new()->Self{
+		Self([SubmeshFaceId(0);2])
+	}
+	fn push(&mut self,i:usize,face_id:SubmeshFaceId){
+		self.0[i]=face_id;
+	}
+}
+struct FaceRefEdges(Vec<SubmeshDirectedEdgeId>);
+#[derive(Default)]
+struct EdgePool{
+	edge_guys:Vec<(EdgeRefVerts,EdgeRefFaces)>,
+	edge_id_from_guy:HashMap<EdgeRefVerts,SubmeshEdgeId>,
+}
+impl EdgePool{
+	fn push(&mut self,edge_ref_verts:EdgeRefVerts)->(&mut EdgeRefFaces,SubmeshEdgeId){
+		let edge_id=*self.edge_id_from_guy.entry(edge_ref_verts.clone()).or_insert_with(||{
+			let edge_id=SubmeshEdgeId::new(self.edge_guys.len() as u32);
+			self.edge_guys.push((edge_ref_verts,EdgeRefFaces::new()));
+			edge_id
+		});
+		(&mut self.edge_guys[edge_id.get() as usize].1,edge_id)
+	}
+}
+
+#[derive(Debug)]
+pub enum PhysicsMeshError{
+	ZeroVertices,
+	NoPhysicsGroups,
+}
+impl std::fmt::Display for PhysicsMeshError{
+	fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+		write!(f,"{self:?}")
+	}
+}
+impl std::error::Error for PhysicsMeshError{}
+
+impl TryFrom<&model::Mesh> for PhysicsMesh{
+	type Error=PhysicsMeshError;
+	fn try_from(mesh:&model::Mesh)->Result<Self,PhysicsMeshError>{
+		if mesh.unique_pos.len()==0{
+			return Err(PhysicsMeshError::ZeroVertices);
+		}
+		// An empty physics mesh is a waste of resources
+		if mesh.physics_groups.len()==0{
+			return Err(PhysicsMeshError::NoPhysicsGroups);
+		}
+		let verts=mesh.unique_pos.iter().copied().map(Vert).collect();
+		//TODO: fix submeshes
+		//flat map mesh.physics_groups[$1].groups.polys()[$2] as face_id
+		//lower face_id points to upper face_id
+		//the same face is not allowed to be in multiple polygon groups
+		// because SubmeshFaceId -> CompleteMeshFaceId -> SubmeshFaceId is ambiguous
+		// when multiple SubmeshFaceId point to one MeshFaceId
+		let mut faces=Vec::new();
+		let mut face_id_from_face=HashMap::new();
+		let mesh_topologies:Vec<PhysicsMeshTopology>=mesh.physics_groups.iter().map(|physics_group|{
+			//construct submesh
+			let mut submesh_faces=Vec::new();//these contain a map from submeshId->meshId
+			let mut submesh_verts=Vec::new();
+			let mut submesh_vert_id_from_mesh_vert_id=HashMap::<MeshVertId,SubmeshVertId>::new();
+			//lazy closure
+			let mut get_submesh_vert_id=|vert_id:MeshVertId|*submesh_vert_id_from_mesh_vert_id.entry(vert_id).or_insert_with(||{
+				let submesh_vert_id=SubmeshVertId::new(submesh_verts.len() as u32);
+				submesh_verts.push(vert_id);
+				submesh_vert_id
+			});
+			let mut edge_pool=EdgePool::default();
+			let mut vert_ref_guys=vec![VertRefGuy::default();mesh.unique_pos.len()];
+			let mut face_ref_guys=Vec::new();
+			for polygon_group_id in &physics_group.groups{
+				let polygon_group=&mesh.polygon_groups[polygon_group_id.get() as usize];
+				for poly_vertices in polygon_group.polys(){
+					let submesh_face_id=SubmeshFaceId::new(submesh_faces.len() as u32);
+					//one face per poly
+					let mut normal=Vector3::new([Fixed::ZERO,Fixed::ZERO,Fixed::ZERO]);
+					let len=poly_vertices.len();
+					let face_edges=poly_vertices.into_iter().enumerate().map(|(i,vert_id)|{
+						let vert0_id=MeshVertId::new(mesh.unique_vertices[vert_id.get() as usize].pos.get() as u32);
+						let vert1_id=MeshVertId::new(mesh.unique_vertices[poly_vertices[(i+1)%len].get() as usize].pos.get() as u32);
+						//index submesh verts
+						let submesh_vert0_id=get_submesh_vert_id(vert0_id);
+						let submesh_vert1_id=get_submesh_vert_id(vert1_id);
+						//https://www.khronos.org/opengl/wiki/Calculating_a_Surface_Normal (Newell's Method)
+						let v0=mesh.unique_pos[vert0_id.get() as usize];
+						let v1=mesh.unique_pos[vert1_id.get() as usize];
+						normal+=Vector3::new([
+							(v0.y-v1.y)*(v0.z+v1.z),
+							(v0.z-v1.z)*(v0.x+v1.x),
+							(v0.x-v1.x)*(v0.y+v1.y),
+						]);
+						//get/create edge and push face into it
+						let (edge_ref_verts,is_sorted)=EdgeRefVerts::new(submesh_vert0_id,submesh_vert1_id);
+						let (edge_ref_faces,edge_id)=edge_pool.push(edge_ref_verts);
+						//polygon vertices as assumed to be listed clockwise
+						//populate the edge face on the left or right depending on how the edge vertices got sorted
+						edge_ref_faces.push(!is_sorted as usize,submesh_face_id);
+						//index edges & face into vertices
+						{
+							let vert_ref_guy=&mut vert_ref_guys[submesh_vert0_id.get() as usize];
+							vert_ref_guy.edges.insert(edge_id.as_directed(is_sorted));
+							vert_ref_guy.faces.insert(submesh_face_id);
+							vert_ref_guys[submesh_vert1_id.get() as usize].edges.insert(edge_id.as_directed(!is_sorted));
+						}
+						//return directed_edge_id
+						edge_id.as_directed(is_sorted)
+					}).collect();
+					let mut dot=Fixed::ZERO;
+					// find the average dot
+					for &v in poly_vertices{
+						dot+=normal.dot(mesh.unique_pos[mesh.unique_vertices[v.get() as usize].pos.get() as usize]);
+					}
+					//assume face hash is stable, and there are no flush faces...
+					let face=Face{
+						normal:(normal/len as i64).divide().narrow_64().unwrap(),
+						dot:(dot/(len*len) as i64).narrow_64().unwrap(),
+					};
+					let face_id=*face_id_from_face.entry(face).or_insert_with(||{
+						let face_id=MeshFaceId::new(faces.len() as u32);
+						faces.push(face);
+						face_id
+					});
+					submesh_faces.push(face_id);
+					face_ref_guys.push(FaceRefEdges(face_edges));
+				}
+			}
+			PhysicsMeshTopology{
+				faces:submesh_faces,
+				verts:submesh_verts,
+				face_topology:face_ref_guys.into_iter().map(|FaceRefEdges(edges)|{
+					FaceRefs{edges}
+				}).collect(),
+				edge_topology:edge_pool.edge_guys.into_iter().map(|(EdgeRefVerts(verts),EdgeRefFaces(faces))|
+					EdgeRefs{faces,verts}
+				).collect(),
+				vert_topology:vert_ref_guys.into_iter().map(|VertRefGuy{edges,faces}|
+					VertRefs{
+						edges:edges.into_iter().collect(),
+						faces:faces.into_iter().collect(),
+					}
+				).collect(),
+			}
+		}).collect();
+		Ok(Self{
+			data:PhysicsMeshData{
+				faces,
+				verts,
+			},
+			// complete_mesh:None,
+			submeshes:mesh_topologies,
+		})
+	}
+}
+
+#[derive(Debug,Clone,Copy)]
+pub struct PhysicsMeshView<'a>{
+	data:&'a PhysicsMeshData,
+	topology:&'a PhysicsMeshTopology,
+}
+impl MeshQuery for PhysicsMeshView<'_>{
+	type Position=Planar64Vec3;
+	type Direction=Planar64Vec3;
+	type Normal=Planar64Vec3;
+	type Offset=Planar64;
+	fn face_nd(&self,face_id:SubmeshFaceId)->(Planar64Vec3,Planar64){
+		let face_idx=self.topology.faces[face_id.get() as usize].get() as usize;
+		(self.data.faces[face_idx].normal,self.data.faces[face_idx].dot)
+	}
+	fn hint_point(&self)->Planar64Vec3{
+		// invariant: meshes always encompass the origin
+		vec3::zero()
+	}
+	fn farthest_vert(&self,dir:Planar64Vec3)->SubmeshVertId{
+		//this happens to be well-defined.  there are no virtual virtices
+		SubmeshVertId::new(
+			self.topology.verts.iter()
+			.enumerate()
+			.max_by_key(|&(_,&vert_id)|
+				dir.dot(self.data.verts[vert_id.get() as usize].0)
+			)
+			//assume there is more than zero vertices.
+			.unwrap().0 as u32
+		)
+	}
+	//ideally I never calculate the vertex position, but I have to for the graphical meshes...
+	fn vert(&self,vert_id:SubmeshVertId)->Planar64Vec3{
+		let vert_idx=self.topology.verts[vert_id.get() as usize].get() as usize;
+		self.data.verts[vert_idx].0
+	}
+	fn edge_n(&self,edge_id:Self::Edge)->Self::Direction{
+		let &[v0,v1]=self.edge_verts(edge_id).as_ref();
+		self.vert(v1)-self.vert(v0)
+	}
+	fn directed_edge_n(&self,directed_edge_id:Self::DirectedEdge)->Self::Direction{
+		let &[v0,v1]=self.edge_verts(directed_edge_id.as_undirected()).as_ref();
+		(self.vert(v1)-self.vert(v0))*((directed_edge_id.parity() as i64)*2-1)
+	}
+}
+impl MeshTopology for PhysicsMeshView<'_>{
+	type Face=SubmeshFaceId;
+	type Edge=SubmeshEdgeId;
+	type DirectedEdge=SubmeshDirectedEdgeId;
+	type Vert=SubmeshVertId;
+	fn for_each_vert_edge(&self,vert_id:Self::Vert,f:impl FnMut(Self::DirectedEdge)){
+		self.topology.vert_topology[vert_id.get() as usize].edges.iter().copied().for_each(f);
+	}
+	fn for_each_vert_face(&self,vert_id:Self::Vert,f:impl FnMut(Self::Face)){
+		self.topology.vert_topology[vert_id.get() as usize].faces.iter().copied().for_each(f);
+	}
+	fn edge_faces(&self,edge_id:Self::Edge)->impl AsRef<[Self::Face;2]>{
+		AsRefHelper(self.topology.edge_topology[edge_id.get() as usize].faces)
+	}
+	fn edge_verts(&self,edge_id:Self::Edge)->impl AsRef<[Self::Vert;2]>{
+		AsRefHelper(self.topology.edge_topology[edge_id.get() as usize].verts)
+	}
+	fn for_each_face_vert(&self,_face_id:Self::Face,_f:impl FnMut(Self::Vert)){
+		unimplemented!()
+	}
+	fn for_each_face_edge(&self,face_id:Self::Face,f:impl FnMut(Self::DirectedEdge)){
+		self.topology.face_topology[face_id.get() as usize].edges.iter().copied().for_each(f);
+	}
+}
+
+#[derive(Debug)]
+pub struct PhysicsMeshTransform{
+	pub vertex:integer::Planar64Affine3,
+	pub normal:integer::mat3::Matrix3<F128_64>,
+	pub det:F192_96,
+}
+impl PhysicsMeshTransform{
+	pub fn new(transform:integer::Planar64Affine3)->Self{
+		Self{
+			normal:transform.matrix3.adjugate().transpose(),
+			det:transform.matrix3.det(),
+			vertex:transform,
+		}
+	}
+}
+
+#[derive(Debug,Clone,Copy)]
+pub struct TransformedMesh<'a>{
+	view:PhysicsMeshView<'a>,
+	transform:&'a PhysicsMeshTransform,
+}
+impl TransformedMesh<'_>{
+	pub const fn new<'a>(
+		view:PhysicsMeshView<'a>,
+		transform:&'a PhysicsMeshTransform,
+	)->TransformedMesh<'a>{
+		TransformedMesh{
+			view,
+			transform,
+		}
+	}
+	pub fn verts<'a>(&'a self)->impl Iterator<Item=Vector3<F128_64>>+'a{
+		self.view.data.verts.iter().map(|&Vert(pos)|self.transform.vertex.transform_point3(pos))
+	}
+}
+impl MeshQuery for TransformedMesh<'_>{
+	type Direction=Planar64Vec3;
+	type Position=Planar64Vec3;
+	type Normal=Vector3<F192_96>;
+	type Offset=F256_128;
+	fn face_nd(&self,face_id:SubmeshFaceId)->(Self::Normal,Self::Offset){
+		let (n,d)=self.view.face_nd(face_id);
+		let transformed_n=self.transform.normal*n;
+		let transformed_d=d*self.transform.det+transformed_n.dot(self.transform.vertex.translation);
+		(transformed_n,transformed_d)
+	}
+	fn vert(&self,vert_id:SubmeshVertId)->Planar64Vec3{
+		// wrap for speed
+		self.transform.vertex.transform_point3(self.view.vert(vert_id)).wrap_64()
+	}
+	fn hint_point(&self)->Planar64Vec3{
+		self.transform.vertex.translation
+	}
+	fn farthest_vert(&self,dir:Planar64Vec3)->SubmeshVertId{
+		//this happens to be well-defined.  there are no virtual virtices
+		SubmeshVertId::new(
+			self.view.topology.verts.iter()
+			.enumerate()
+			.max_by_key(|&(_,&vert_id)|
+				dir.dot(self.transform.vertex.transform_point3(self.view.data.verts[vert_id.get() as usize].0))
+			)
+			//assume there is more than zero vertices.
+			.unwrap().0 as u32
+		)
+	}
+	fn edge_n(&self,edge_id:Self::Edge)->Self::Direction{
+		let &[v0,v1]=self.edge_verts(edge_id).as_ref();
+		self.vert(v1)-self.vert(v0)
+	}
+	fn directed_edge_n(&self,directed_edge_id:Self::DirectedEdge)->Self::Direction{
+		let &[v0,v1]=self.edge_verts(directed_edge_id.as_undirected()).as_ref();
+		(self.vert(v1)-self.vert(v0))*((directed_edge_id.parity() as i64)*2-1)
+	}
+}
+impl MeshTopology for TransformedMesh<'_>{
+	type Face=SubmeshFaceId;
+	type Edge=SubmeshEdgeId;
+	type DirectedEdge=SubmeshDirectedEdgeId;
+	type Vert=SubmeshVertId;
+	#[inline]
+	fn for_each_vert_edge(&self,vert_id:Self::Vert,f:impl FnMut(Self::DirectedEdge)){
+		self.view.for_each_vert_edge(vert_id,f)
+	}
+	#[inline]
+	fn for_each_vert_face(&self,vert_id:Self::Vert,f:impl FnMut(Self::Face)){
+		self.view.for_each_vert_face(vert_id,f)
+	}
+	#[inline]
+	fn edge_faces(&self,edge_id:Self::Edge)->impl AsRef<[Self::Face;2]>{
+		self.view.edge_faces(edge_id)
+	}
+	#[inline]
+	fn edge_verts(&self,edge_id:Self::Edge)->impl AsRef<[Self::Vert;2]>{
+		self.view.edge_verts(edge_id)
+	}
+	fn for_each_face_vert(&self,_face_id:Self::Face,_f:impl FnMut(Self::Vert)){
+		unimplemented!()
+	}
+	#[inline]
+	fn for_each_face_edge(&self,face_id:Self::Face,f:impl FnMut(Self::DirectedEdge)){
+		self.view.for_each_face_edge(face_id,f)
+	}
+}
+
+pub type GigaTime=Ratio<F256_128,F256_128>;
+pub fn into_giga_time(time:Time,relative_to:Time)->GigaTime{
+	let r=(time-relative_to).to_ratio();
+	Ratio::new(r.num.widen_256(),r.den.widen_256())
+}

engine/physics/src/push_solve.rs

@@ -1,4 +1,7 @@
-use strafesnet_common::integer::{self,vec3::{self,Vector3},Fixed,Planar64,Planar64Vec3,Ratio};
+use strafesnet_common::integer::fixed_types::{F64_32,F128_64,F192_96,F256_128,F320_160};
+use strafesnet_common::integer::vec3::{self,Vector3};
+use strafesnet_common::integer::{Fixed,Planar64Vec3,Ratio};
+use strafesnet_common::ray::Ray;
 
 // This algorithm is based on Lua code
 // written by Trey Reynolds in 2021
@@ -10,27 +13,10 @@ use strafesnet_common::integer::{self,vec3::{self,Vector3},Fixed,Planar64,Planar
 type Conts<'a>=arrayvec::ArrayVec<&'a Contact,4>;
 
 // hack to allow comparing ratios to zero
-const RATIO_ZERO:Ratio<Fixed<1,32>,Fixed<1,32>>=Ratio::new(Fixed::ZERO,Fixed::EPSILON);
-
-struct Ray{
-	origin:Planar64Vec3,
-	direction:Planar64Vec3,
-}
-impl Ray{
-	fn extrapolate<Num,Den,N1,T1>(&self,t:Ratio<Num,Den>)->Planar64Vec3
-	where
-		Num:Copy,
-		Den:Copy,
-		Num:core::ops::Mul<Planar64,Output=N1>,
-		Planar64:core::ops::Mul<Den,Output=N1>,
-		N1:integer::Divide<Den,Output=T1>,
-		T1:integer::Fix<Planar64>,
-	{
-		self.origin+self.direction.map(|elem|(t*elem).divide().fix())
-	}
-}
+const RATIO_ZERO:Ratio<F64_32,F64_32>=Ratio::new(Fixed::ZERO,Fixed::EPSILON);
 
 /// Information about a contact restriction
+#[derive(Debug,PartialEq)]
 pub struct Contact{
 	pub position:Planar64Vec3,
 	pub velocity:Planar64Vec3,
@@ -44,42 +30,39 @@ impl Contact{
 			normal:self.normal,
 		}
 	}
-	fn relative_dot(&self,direction:Planar64Vec3)->Fixed<2,64>{
+	fn relative_dot(&self,direction:Planar64Vec3)->F128_64{
 		(direction-self.velocity).dot(self.normal)
 	}
 	/// Calculate the time of intersection. (previously get_touch_time)
-	fn solve(&self,ray:&Ray)->Ratio<Fixed<2,64>,Fixed<2,64>>{
+	fn solve(&self,ray:&Ray)->Ratio<F128_64,F128_64>{
 		(self.position-ray.origin).dot(self.normal)/(ray.direction-self.velocity).dot(self.normal)
 	}
 }
 
 //note that this is horrible with fixed point arithmetic
-fn solve1(c0:&Contact)->Option<Ratio<Vector3<Fixed<3,96>>,Fixed<2,64>>>{
-	const EPSILON:Fixed<2,64>=Fixed::from_bits(Fixed::<2,64>::ONE.to_bits().shr(10));
+fn solve1(c0:&Contact)->Option<Ratio<Vector3<F192_96>,F128_64>>{
 	let det=c0.normal.dot(c0.velocity);
-	if det.abs()<EPSILON{
+	if det.abs()==Fixed::ZERO{
 		return None;
 	}
 	let d0=c0.normal.dot(c0.position);
 	Some(c0.normal*d0/det)
 }
-fn solve2(c0:&Contact,c1:&Contact)->Option<Ratio<Vector3<Fixed<5,160>>,Fixed<4,128>>>{
-	const EPSILON:Fixed<4,128>=Fixed::from_bits(Fixed::<4,128>::ONE.to_bits().shr(10));
+fn solve2(c0:&Contact,c1:&Contact)->Option<Ratio<Vector3<F320_160>,F256_128>>{
 	let u0_u1=c0.velocity.cross(c1.velocity);
 	let n0_n1=c0.normal.cross(c1.normal);
 	let det=u0_u1.dot(n0_n1);
-	if det.abs()<EPSILON{
+	if det.abs()==Fixed::ZERO{
 		return None;
 	}
 	let d0=c0.normal.dot(c0.position);
 	let d1=c1.normal.dot(c1.position);
 	Some((c1.normal.cross(u0_u1)*d0+u0_u1.cross(c0.normal)*d1)/det)
 }
-fn solve3(c0:&Contact,c1:&Contact,c2:&Contact)->Option<Ratio<Vector3<Fixed<4,128>>,Fixed<3,96>>>{
-	const EPSILON:Fixed<3,96>=Fixed::from_bits(Fixed::<3,96>::ONE.to_bits().shr(10));
+fn solve3(c0:&Contact,c1:&Contact,c2:&Contact)->Option<Ratio<Vector3<F256_128>,F192_96>>{
 	let n0_n1=c0.normal.cross(c1.normal);
 	let det=c2.normal.dot(n0_n1);
-	if det.abs()<EPSILON{
+	if det.abs()==Fixed::ZERO{
 		return None;
 	}
 	let d0=c0.normal.dot(c0.position);
@@ -88,7 +71,7 @@ fn solve3(c0:&Contact,c1:&Contact,c2:&Contact)->Option<Ratio<Vector3<Fixed<4,128
 	Some((c1.normal.cross(c2.normal)*d0+c2.normal.cross(c0.normal)*d1+c0.normal.cross(c1.normal)*d2)/det)
 }
 
-fn decompose1(point:Planar64Vec3,u0:Planar64Vec3)->Option<[Ratio<Fixed<2,64>,Fixed<2,64>>;1]>{
+fn decompose1(point:Planar64Vec3,u0:Planar64Vec3)->Option<[Ratio<F128_64,F128_64>;1]>{
 	let det=u0.dot(u0);
 	if det==Fixed::ZERO{
 		return None;
@@ -96,7 +79,7 @@ fn decompose1(point:Planar64Vec3,u0:Planar64Vec3)->Option<[Ratio<Fixed<2,64>,Fix
 	let s0=u0.dot(point)/det;
 	Some([s0])
 }
-fn decompose2(point:Planar64Vec3,u0:Planar64Vec3,u1:Planar64Vec3)->Option<[Ratio<Fixed<4,128>,Fixed<4,128>>;2]>{
+fn decompose2(point:Planar64Vec3,u0:Planar64Vec3,u1:Planar64Vec3)->Option<[Ratio<F256_128,F256_128>;2]>{
 	let u0_u1=u0.cross(u1);
 	let det=u0_u1.dot(u0_u1);
 	if det==Fixed::ZERO{
@@ -106,7 +89,7 @@ fn decompose2(point:Planar64Vec3,u0:Planar64Vec3,u1:Planar64Vec3)->Option<[Ratio
 	let s1=u0_u1.dot(u0.cross(point))/det;
 	Some([s0,s1])
 }
-fn decompose3(point:Planar64Vec3,u0:Planar64Vec3,u1:Planar64Vec3,u2:Planar64Vec3)->Option<[Ratio<Fixed<3,96>,Fixed<3,96>>;3]>{
+fn decompose3(point:Planar64Vec3,u0:Planar64Vec3,u1:Planar64Vec3,u2:Planar64Vec3)->Option<[Ratio<F192_96,F192_96>;3]>{
 	let det=u0.cross(u1).dot(u2);
 	if det==Fixed::ZERO{
 		return None;
@@ -165,21 +148,22 @@ fn is_space_enclosed_4(
 }
 
 const fn get_push_ray_0(point:Planar64Vec3)->Ray{
-	Ray{origin:point,direction:vec3::ZERO}
+	Ray{origin:point,direction:vec3::zero()}
 }
 fn get_push_ray_1(point:Planar64Vec3,c0:&Contact)->Option<Ray>{
-	let direction=solve1(c0)?.divide().fix_1();
+	//wrap for speed
+	let direction=solve1(c0)?.divide().wrap_64();
 	let [s0]=decompose1(direction,c0.velocity)?;
 	if s0.lt_ratio(RATIO_ZERO){
 		return None;
 	}
 	let origin=point+solve1(
 		&c0.relative_to(point),
-	)?.divide().fix_1();
+	)?.divide().wrap_64();
 	Some(Ray{origin,direction})
 }
 fn get_push_ray_2(point:Planar64Vec3,c0:&Contact,c1:&Contact)->Option<Ray>{
-	let direction=solve2(c0,c1)?.divide().fix_1();
+	let direction=solve2(c0,c1)?.divide().wrap_64();
 	let [s0,s1]=decompose2(direction,c0.velocity,c1.velocity)?;
 	if s0.lt_ratio(RATIO_ZERO)||s1.lt_ratio(RATIO_ZERO){
 		return None;
@@ -187,11 +171,11 @@ fn get_push_ray_2(point:Planar64Vec3,c0:&Contact,c1:&Contact)->Option<Ray>{
 	let origin=point+solve2(
 		&c0.relative_to(point),
 		&c1.relative_to(point),
-	)?.divide().fix_1();
+	)?.divide().wrap_64();
 	Some(Ray{origin,direction})
 }
 fn get_push_ray_3(point:Planar64Vec3,c0:&Contact,c1:&Contact,c2:&Contact)->Option<Ray>{
-	let direction=solve3(c0,c1,c2)?.divide().fix_1();
+	let direction=solve3(c0,c1,c2)?.divide().wrap_64();
 	let [s0,s1,s2]=decompose3(direction,c0.velocity,c1.velocity,c2.velocity)?;
 	if s0.lt_ratio(RATIO_ZERO)||s1.lt_ratio(RATIO_ZERO)||s2.lt_ratio(RATIO_ZERO){
 		return None;
@@ -200,14 +184,14 @@ fn get_push_ray_3(point:Planar64Vec3,c0:&Contact,c1:&Contact,c2:&Contact)->Optio
 		&c0.relative_to(point),
 		&c1.relative_to(point),
 		&c2.relative_to(point),
-	)?.divide().fix_1();
+	)?.divide().wrap_64();
 	Some(Ray{origin,direction})
 }
 
 const fn get_best_push_ray_and_conts_0<'a>(point:Planar64Vec3)->(Ray,Conts<'a>){
 	(get_push_ray_0(point),Conts::new_const())
 }
-fn get_best_push_ray_and_conts_1(point:Planar64Vec3,c0:&Contact)->Option<(Ray,Conts)>{
+fn get_best_push_ray_and_conts_1(point:Planar64Vec3,c0:&Contact)->Option<(Ray,Conts<'_>)>{
 	get_push_ray_1(point,c0)
 		.map(|ray|(ray,Conts::from_iter([c0])))
 }
@@ -289,26 +273,26 @@ fn get_best_push_ray_and_conts<'a>(
 	}
 }
 
-fn get_first_touch<'a>(contacts:&'a Vec<Contact>,ray:&Ray,conts:&Conts)->Option<(Ratio<Fixed<2,64>,Fixed<2,64>>,&'a Contact)>{
+fn get_first_touch<'a>(contacts:&'a [Contact],ray:&Ray,conts:&Conts)->Option<(Ratio<F128_64,F128_64>,&'a Contact)>{
 	contacts.iter()
 		.filter(|&contact|
-			!conts.iter().any(|&c|std::ptr::eq(c,contact))
+			!conts.iter().any(|&c|core::ptr::eq(c,contact))
 			&&contact.relative_dot(ray.direction).is_negative()
 		)
 		.map(|contact|(contact.solve(ray),contact))
 		.min_by_key(|&(t,_)|t)
 }
 
-pub fn push_solve(contacts:&Vec<Contact>,point:Planar64Vec3)->Planar64Vec3{
+pub fn push_solve(contacts:&[Contact],point:Planar64Vec3)->(Planar64Vec3,Conts<'_>){
 	let (mut ray,mut conts)=get_best_push_ray_and_conts_0(point);
 	loop{
 		let (next_t,next_cont)=match get_first_touch(contacts,&ray,&conts){
 			Some((t,cont))=>(t,cont),
-			None=>return ray.origin,
+			None=>return (ray.origin,conts),
 		};
 
 		if RATIO_ZERO.le_ratio(next_t){
-			return ray.origin;
+			return (ray.origin,conts);
 		}
 
 		//push_front
@@ -324,7 +308,7 @@ pub fn push_solve(contacts:&Vec<Contact>,point:Planar64Vec3)->Planar64Vec3{
 		let meet_point=ray.extrapolate(next_t);
 		match get_best_push_ray_and_conts(meet_point,conts.as_slice()){
 			Some((new_ray,new_conts))=>(ray,conts)=(new_ray,new_conts),
-			None=>return meet_point,
+			None=>return (meet_point,conts),
 		}
 	}
 }
@@ -336,14 +320,13 @@ mod tests{
 	fn test_push_solve(){
 		let contacts=vec![
 			Contact{
-				position:vec3::ZERO,
+				position:vec3::zero(),
 				velocity:vec3::Y,
 				normal:vec3::Y,
 			}
 		];
-		assert_eq!(
-			vec3::ZERO,
-			push_solve(&contacts,vec3::NEG_Y)
-		);
+		let (point,conts)=push_solve(&contacts,vec3::NEG_Y);
+		assert_eq!(point,vec3::zero());
+		assert_eq!(conts.as_slice(),[&contacts[0]].as_slice());
 	}
 }

engine/session/Cargo.toml

@@ -0,0 +1,15 @@
+[package]
+name = "strafesnet_session"
+version = "0.1.0"
+edition = "2024"
+
+[dependencies]
+glam.workspace = true
+replace_with = "0.1.7"
+strafesnet_common.workspace = true
+strafesnet_physics.workspace = true
+strafesnet_settings.workspace = true
+strafesnet_snf.workspace = true
+
+[lints]
+workspace = true

engine/session/LICENSE

@@ -0,0 +1,8 @@
+/*******************************************************
+* Copyright (C) 2023-2024 Rhys Lloyd <krakow20@gmail.com>
+*
+* This file is part of the StrafesNET bhop/surf client.
+*
+* StrafesNET can not be copied and/or distributed
+* without the express permission of Rhys Lloyd
+*******************************************************/

engine/session/src/lib.rs

@@ -0,0 +1,2 @@
+mod mouse_interpolator;
+pub mod session;

engine/session/src/mouse_interpolator.rs

@@ -5,13 +5,13 @@ use strafesnet_common::physics::{
 	TimeInner as PhysicsTimeInner,
 	Time as PhysicsTime,
 };
-use strafesnet_common::session::{Time as SessionTime,TimeInner as SessionTimeInner};
+use strafesnet_common::session::Time as SessionTime;
 use strafesnet_common::instruction::{InstructionConsumer,InstructionEmitter,TimedInstruction};
 
-type TimedSelfInstruction=TimedInstruction<Instruction,PhysicsTimeInner>;
-type DoubleTimedSelfInstruction=TimedInstruction<TimedSelfInstruction,SessionTimeInner>;
+type TimedSelfInstruction=TimedInstruction<Instruction,PhysicsTime>;
+type DoubleTimedSelfInstruction=TimedInstruction<TimedSelfInstruction,SessionTime>;
 
-type TimedPhysicsInstruction=TimedInstruction<PhysicsInstruction,PhysicsTimeInner>;
+type TimedPhysicsInstruction=TimedInstruction<PhysicsInstruction,PhysicsTime>;
 
 const MOUSE_TIMEOUT:SessionTime=SessionTime::from_millis(10);
 
@@ -89,14 +89,14 @@ pub struct MouseInterpolator{
 // Maybe MouseInterpolator manipulation is better expressed using impls
 // and called from Instruction trait impls in session
 impl InstructionConsumer<TimedSelfInstruction> for MouseInterpolator{
-	type TimeInner=SessionTimeInner;
+	type Time=SessionTime;
 	fn process_instruction(&mut self,ins:DoubleTimedSelfInstruction){
 		self.push_unbuffered_input(ins.time,ins.instruction.time,ins.instruction.instruction.into())
 	}
 }
 impl InstructionEmitter<StepInstruction> for MouseInterpolator{
-	type TimeInner=SessionTimeInner;
-	fn next_instruction(&self,time_limit:SessionTime)->Option<TimedInstruction<StepInstruction,Self::TimeInner>>{
+	type Time=SessionTime;
+	fn next_instruction(&self,time_limit:SessionTime)->Option<TimedInstruction<StepInstruction,Self::Time>>{
 		self.buffered_instruction_with_timeout(time_limit)
 	}
 }
@@ -108,7 +108,7 @@ impl MouseInterpolator{
 			output:std::collections::VecDeque::new(),
 		}
 	}
-	fn push_mouse_and_flush_buffer(&mut self,ins:TimedInstruction<MouseInstruction,PhysicsTimeInner>){
+	fn push_mouse_and_flush_buffer(&mut self,ins:TimedInstruction<MouseInstruction,PhysicsTime>){
 		self.buffer.push_front(TimedInstruction{
 			time:ins.time,
 			instruction:BufferedInstruction::Mouse(ins.instruction).into(),
@@ -219,7 +219,7 @@ impl MouseInterpolator{
 			}
 		}
 	}
-	fn buffered_instruction_with_timeout(&self,time_limit:SessionTime)->Option<TimedInstruction<StepInstruction,SessionTimeInner>>{
+	fn buffered_instruction_with_timeout(&self,time_limit:SessionTime)->Option<TimedInstruction<StepInstruction,SessionTime>>{
 		match self.get_mouse_timedout_at(time_limit){
 			Some(timeout)=>Some(TimedInstruction{
 				time:timeout,
@@ -232,7 +232,7 @@ impl MouseInterpolator{
 			}),
 		}
 	}
-	pub fn pop_buffered_instruction(&mut self,ins:TimedInstruction<StepInstruction,PhysicsTimeInner>)->Option<TimedPhysicsInstruction>{
+	pub fn pop_buffered_instruction(&mut self,ins:TimedInstruction<StepInstruction,PhysicsTime>)->Option<TimedPhysicsInstruction>{
 		match ins.instruction{
 			StepInstruction::Pop=>(),
 			StepInstruction::Timeout=>self.timeout_mouse(ins.time),
@@ -244,6 +244,7 @@ impl MouseInterpolator{
 #[cfg(test)]
 mod test{
 	use super::*;
+	use strafesnet_common::session::TimeInner as SessionTimeInner;
 	#[test]
 	fn test(){
 		let mut interpolator=MouseInterpolator::new();

engine/session/src/session.rs

@@ -12,10 +12,11 @@ use strafesnet_common::physics::{
 };
 use strafesnet_common::timer::{Scaled,Timer};
 use strafesnet_common::session::{TimeInner as SessionTimeInner,Time as SessionTime};
+use strafesnet_settings::directories::Directories;
 
 use crate::mouse_interpolator::{MouseInterpolator,StepInstruction,Instruction as MouseInterpolatorInstruction};
-use crate::physics::{PhysicsContext,PhysicsData};
-use crate::settings::UserSettings;
+use strafesnet_physics::physics::{self,PhysicsContext,PhysicsData};
+use strafesnet_settings::settings::UserSettings;
 
 pub enum Instruction<'a>{
 	Input(SessionInputInstruction),
@@ -30,7 +31,7 @@ pub enum SessionInputInstruction{
 	Mouse(glam::IVec2),
 	SetControl(strafesnet_common::physics::SetControlInstruction),
 	Mode(ImplicitModeInstruction),
-	Misc(strafesnet_common::physics::MiscInstruction),
+	Misc(MiscInstruction),
 }
 /// Implicit mode instruction are fed separately to session.
 /// Session generates the explicit mode instructions interlaced with a SetSensitivity instruction
@@ -51,34 +52,41 @@ pub enum SessionControlInstruction{
 pub enum SessionPlaybackInstruction{
 	SkipForward,
 	SkipBack,
-	TogglePaused,
 	DecreaseTimescale,
 	IncreaseTimescale,
 }
 
 pub struct FrameState{
-	pub body:crate::physics::Body,
-	pub camera:crate::physics::PhysicsCamera,
+	pub trajectory:physics::Trajectory,
+	pub camera:physics::PhysicsCamera,
 	pub time:PhysicsTime,
 }
+impl FrameState{
+	pub fn pos(&self)->glam::Vec3{
+		self.trajectory.extrapolated_position(self.time).map(Into::<f32>::into).to_array().into()
+	}
+	pub fn angles(&self)->glam::Vec2{
+		self.camera.simulate_move_angles(glam::IVec2::ZERO)
+	}
+}
 
 pub struct Simulation{
 	timer:Timer<Scaled<SessionTimeInner,PhysicsTimeInner>>,
-	physics:crate::physics::PhysicsState,
+	physics:physics::PhysicsState,
 }
 impl Simulation{
 	pub const fn new(
 		timer:Timer<Scaled<SessionTimeInner,PhysicsTimeInner>>,
-		physics:crate::physics::PhysicsState,
+		physics:physics::PhysicsState,
 	)->Self{
 		Self{
 			timer,
 			physics,
 		}
 	}
-	pub fn get_frame_state(&self,time:SessionTime)->FrameState{
+	pub fn get_frame_state(&self,time:SessionTime,data:&PhysicsData)->FrameState{
 		FrameState{
-			body:self.physics.camera_body(),
+			trajectory:self.physics.camera_trajectory(data),
 			camera:self.physics.camera(),
 			time:self.timer.time(time),
 		}
@@ -87,11 +95,11 @@ impl Simulation{
 
 #[derive(Default)]
 pub struct Recording{
-	instructions:Vec<TimedInstruction<PhysicsInputInstruction,PhysicsTimeInner>>,
+	instructions:Vec<TimedInstruction<PhysicsInputInstruction,PhysicsTime>>,
 }
 impl Recording{
-	fn new(
-		instructions:Vec<TimedInstruction<PhysicsInputInstruction,PhysicsTimeInner>>,
+	pub fn new(
+		instructions:Vec<TimedInstruction<PhysicsInputInstruction,PhysicsTime>>,
 	)->Self{
 		Self{instructions}
 	}
@@ -149,11 +157,12 @@ enum ViewState{
 }
 
 pub struct Session{
+	directories:Directories,
 	user_settings:UserSettings,
-	mouse_interpolator:crate::mouse_interpolator::MouseInterpolator,
+	mouse_interpolator:MouseInterpolator,
 	view_state:ViewState,
 	//gui:GuiState
-	geometry_shared:crate::physics::PhysicsData,
+	geometry_shared:PhysicsData,
 	simulation:Simulation,
 	// below fields not included in lite session
 	recording:Recording,
@@ -163,12 +172,14 @@ pub struct Session{
 impl Session{
 	pub fn new(
 		user_settings:UserSettings,
+		directories:Directories,
 		simulation:Simulation,
 	)->Self{
 		Self{
 			user_settings,
+			directories,
 			mouse_interpolator:MouseInterpolator::new(),
-			geometry_shared:Default::default(),
+			geometry_shared:PhysicsData::empty(),
 			simulation,
 			view_state:ViewState::Play,
 			recording:Default::default(),
@@ -180,13 +191,13 @@ impl Session{
 	}
 	fn change_map(&mut self,map:&strafesnet_common::map::CompleteMap){
 		self.simulation.physics.clear();
-		self.geometry_shared.generate_models(map);
+		self.geometry_shared=PhysicsData::new(map);
 	}
 	pub fn get_frame_state(&self,time:SessionTime)->Option<FrameState>{
 		match &self.view_state{
-			ViewState::Play=>Some(self.simulation.get_frame_state(time)),
+			ViewState::Play=>Some(self.simulation.get_frame_state(time,&self.geometry_shared)),
 			ViewState::Replay(bot_id)=>self.replays.get(bot_id).map(|replay|
-				replay.simulation.get_frame_state(time)
+				replay.simulation.get_frame_state(time,&self.geometry_shared)
 			),
 		}
 	}
@@ -203,8 +214,8 @@ impl Session{
 // Session emits DoStep
 
 impl InstructionConsumer<Instruction<'_>> for Session{
-	type TimeInner=SessionTimeInner;
-	fn process_instruction(&mut self,ins:TimedInstruction<Instruction,Self::TimeInner>){
+	type Time=SessionTime;
+	fn process_instruction(&mut self,ins:TimedInstruction<Instruction,Self::Time>){
 		// repetitive procedure macro
 		macro_rules! run_mouse_interpolator_instruction{
 			($instruction:expr)=>{
@@ -235,12 +246,16 @@ impl InstructionConsumer<Instruction<'_>> for Session{
 				run_mouse_interpolator_instruction!(MouseInterpolatorInstruction::Mode(ModeInstruction::Reset));
 				run_mouse_interpolator_instruction!(MouseInterpolatorInstruction::Misc(MiscInstruction::SetSensitivity(self.user_settings().calculate_sensitivity())));
 				run_mouse_interpolator_instruction!(MouseInterpolatorInstruction::Mode(ModeInstruction::Restart(mode_id)));
+				// TODO: think about this harder.  This works around a bug where you fall infinitely when you reset.
+				self.simulation.timer.set_time(ins.time,PhysicsTime::ZERO);
 			},
 			Instruction::Input(SessionInputInstruction::Mode(ImplicitModeInstruction::ResetAndSpawn(mode_id,spawn_id)))=>{
 				self.clear_recording();
 				run_mouse_interpolator_instruction!(MouseInterpolatorInstruction::Mode(ModeInstruction::Reset));
 				run_mouse_interpolator_instruction!(MouseInterpolatorInstruction::Misc(MiscInstruction::SetSensitivity(self.user_settings().calculate_sensitivity())));
 				run_mouse_interpolator_instruction!(MouseInterpolatorInstruction::Mode(ModeInstruction::Spawn(mode_id,spawn_id)));
+				// TODO: think about this harder.  This works around a bug where you fall infinitely when you reset.
+				self.simulation.timer.set_time(ins.time,PhysicsTime::ZERO);
 			},
 			Instruction::Input(SessionInputInstruction::Misc(misc_instruction))=>{
 				run_mouse_interpolator_instruction!(MouseInterpolatorInstruction::Misc(misc_instruction));
@@ -249,7 +264,14 @@ impl InstructionConsumer<Instruction<'_>> for Session{
 				// don't flush the buffered instructions in the mouse interpolator
 				// until the mouse is confirmed to be not moving at a later time
 				// what if they pause for 5ms lmao
-				_=self.simulation.timer.set_paused(ins.time,paused);
+				match &self.view_state{
+					ViewState::Play=>{
+						_=self.simulation.timer.set_paused(ins.time,paused);
+					},
+					ViewState::Replay(bot_id)=>if let Some(replay)=self.replays.get_mut(bot_id){
+						_=replay.simulation.timer.set_paused(ins.time,paused);
+					},
+				}
 			},
 			Instruction::Control(SessionControlInstruction::CopyRecordingIntoReplayAndSpectate)=> if let ViewState::Play=self.view_state{
 				// Bind: B
@@ -292,11 +314,22 @@ impl InstructionConsumer<Instruction<'_>> for Session{
 			Instruction::Control(SessionControlInstruction::SaveReplay)=>{
 				// Bind: N
 				let view_state=core::mem::replace(&mut self.view_state,ViewState::Play);
-				let file=std::fs::File::create(format!("{}.snfb",ins.time)).unwrap();
 				match view_state{
 					ViewState::Play=>(),
 					ViewState::Replay(bot_id)=>if let Some(replay)=self.replays.remove(&bot_id){
-						strafesnet_snf::bot::write_bot(std::io::BufWriter::new(file),replay.recording.instructions).unwrap();
+						let mut replays_path=self.directories.replays.clone();
+						let file_name=format!("{}.snfb",ins.time);
+						std::thread::spawn(move ||{
+							std::fs::create_dir_all(replays_path.as_path()).unwrap();
+							replays_path.push(file_name);
+							let file=std::fs::File::create(replays_path).unwrap();
+							strafesnet_snf::bot::write_bot(
+								std::io::BufWriter::new(file),
+								strafesnet_physics::VERSION.get(),
+								replay.recording.instructions
+							).unwrap();
+							println!("Finished writing bot file!");
+						});
 					},
 				}
 				_=self.simulation.timer.set_paused(ins.time,false);
@@ -359,14 +392,6 @@ impl InstructionConsumer<Instruction<'_>> for Session{
 					},
 				}
 			},
-			Instruction::Playback(SessionPlaybackInstruction::TogglePaused)=>{
-				match &self.view_state{
-					ViewState::Play=>(),
-					ViewState::Replay(bot_id)=>if let Some(replay)=self.replays.get_mut(bot_id){
-						_=replay.simulation.timer.set_paused(ins.time,!replay.simulation.timer.is_paused());
-					},
-				}
-			}
 			Instruction::ChangeMap(complete_map)=>{
 				self.clear_recording();
 				self.change_map(complete_map);
@@ -410,8 +435,8 @@ impl InstructionConsumer<Instruction<'_>> for Session{
 	}
 }
 impl InstructionConsumer<StepInstruction> for Session{
-	type TimeInner=SessionTimeInner;
-	fn process_instruction(&mut self,ins:TimedInstruction<StepInstruction,Self::TimeInner>){
+	type Time=SessionTime;
+	fn process_instruction(&mut self,ins:TimedInstruction<StepInstruction,Self::Time>){
 		let time=self.simulation.timer.time(ins.time);
 		if let Some(instruction)=self.mouse_interpolator.pop_buffered_instruction(ins.set_time(time)){
 			//record
@@ -421,8 +446,8 @@ impl InstructionConsumer<StepInstruction> for Session{
 	}
 }
 impl InstructionEmitter<StepInstruction> for Session{
-	type TimeInner=SessionTimeInner;
-	fn next_instruction(&self,time_limit:SessionTime)->Option<TimedInstruction<StepInstruction,Self::TimeInner>>{
+	type Time=SessionTime;
+	fn next_instruction(&self,time_limit:SessionTime)->Option<TimedInstruction<StepInstruction,Self::Time>>{
 		self.mouse_interpolator.next_instruction(time_limit)
 	}
 }

engine/settings/Cargo.toml

@@ -0,0 +1,13 @@
+[package]
+name = "strafesnet_settings"
+version = "0.1.0"
+edition = "2024"
+
+[dependencies]
+configparser = "3.0.2"
+directories = "6.0.0"
+glam.workspace = true
+strafesnet_common.workspace = true
+
+[lints]
+workspace = true

engine/settings/LICENSE

@@ -0,0 +1,8 @@
+/*******************************************************
+* Copyright (C) 2023-2024 Rhys Lloyd <krakow20@gmail.com>
+*
+* This file is part of the StrafesNET bhop/surf client.
+*
+* StrafesNET can not be copied and/or distributed
+* without the express permission of Rhys Lloyd
+*******************************************************/

engine/settings/src/directories.rs

@@ -0,0 +1,32 @@
+use std::path::PathBuf;
+
+use crate::settings::{UserSettings,load_user_settings};
+
+pub struct Directories{
+	pub settings:PathBuf,
+	pub maps:PathBuf,
+	pub replays:PathBuf,
+}
+impl Directories{
+	pub fn settings(&self)->UserSettings{
+		load_user_settings(&self.settings)
+	}
+	pub fn user()->Option<Self>{
+		let dirs=directories::ProjectDirs::from("net.strafes","StrafesNET","Strafe Client")?;
+		Some(Self{
+			settings:dirs.config_dir().join("settings.conf"),
+			maps:dirs.cache_dir().join("maps"),
+			// separate directory for remote downloaded replays (cache)
+			// bots:dirs.cache_dir().join("bots"),
+			replays:dirs.data_local_dir().join("replays"),
+		})
+	}
+	pub fn portable()->Result<Self,std::io::Error>{
+		let current_dir=std::env::current_dir()?;
+		Ok(Self{
+			settings:current_dir.join("settings.conf"),
+			maps:current_dir.join("maps"),
+			replays:current_dir.join("replays"),
+		})
+	}
+}

engine/settings/src/lib.rs

@@ -0,0 +1,2 @@
+pub mod settings;
+pub mod directories;

engine/settings/src/settings.rs

@@ -74,9 +74,9 @@ sensitivity_y_from_x_ratio=1
 Sensitivity::DeriveY{x:0.0.001,y:DerivedSensitivity{ratio:1.0}}
 */
 
-pub fn read_user_settings()->UserSettings{
+pub fn load_user_settings(path:&std::path::Path)->UserSettings{
 	let mut cfg=configparser::ini::Ini::new();
-	if let Ok(_)=cfg.load("settings.conf"){
+	if let Ok(_)=cfg.load(path){
 		let (cfg_fov_x,cfg_fov_y)=(cfg.getfloat("camera","fov_x"),cfg.getfloat("camera","fov_y"));
 		let fov=match(cfg_fov_x,cfg_fov_y){
 			(Ok(Some(fov_x)),Ok(Some(fov_y)))=>Fov::Exactly {
@@ -136,4 +136,4 @@ pub fn read_user_settings()->UserSettings{
 	}else{
 		UserSettings::default()
 	}
-}
+}

integration-testing/.gitignore

@@ -0,0 +1 @@
+/test_files

integration-testing/Cargo.toml

@@ -0,0 +1,15 @@
+[package]
+name = "integration-testing"
+version = "0.1.0"
+edition = "2024"
+
+[dependencies]
+glam.workspace = true
+strafesnet_common.workspace = true
+strafesnet_physics.workspace = true
+strafesnet_snf.workspace = true
+# this is just for the primitive constructor
+strafesnet_rbx_loader.workspace = true
+
+[lints]
+workspace = true

integration-testing/src/error.rs

@@ -0,0 +1,28 @@
+#[expect(dead_code)]
+#[derive(Debug)]
+pub enum ReplayError{
+	IO(std::io::Error),
+	SNF(strafesnet_snf::Error),
+	SNFM(strafesnet_snf::map::Error),
+	SNFB(strafesnet_snf::bot::Error),
+}
+impl From<std::io::Error> for ReplayError{
+	fn from(value:std::io::Error)->Self{
+		Self::IO(value)
+	}
+}
+impl From<strafesnet_snf::Error> for ReplayError{
+	fn from(value:strafesnet_snf::Error)->Self{
+		Self::SNF(value)
+	}
+}
+impl From<strafesnet_snf::map::Error> for ReplayError{
+	fn from(value:strafesnet_snf::map::Error)->Self{
+		Self::SNFM(value)
+	}
+}
+impl From<strafesnet_snf::bot::Error> for ReplayError{
+	fn from(value:strafesnet_snf::bot::Error)->Self{
+		Self::SNFB(value)
+	}
+}

integration-testing/src/main.rs

@@ -0,0 +1,225 @@
+mod error;
+mod util;
+
+#[cfg(test)]
+mod tests;
+#[cfg(test)]
+mod test_scenes;
+
+use std::time::Instant;
+
+use error::ReplayError;
+use util::read_entire_file;
+use strafesnet_physics::physics::{PhysicsData,PhysicsState,PhysicsContext};
+
+fn main(){
+	let arg=std::env::args().skip(1).next();
+	match arg.as_deref(){
+		Some("determinism")|None=>test_determinism().unwrap(),
+		Some("replay")=>run_replay().unwrap(),
+		_=>println!("invalid argument"),
+	}
+}
+
+fn run_replay()->Result<(),ReplayError>{
+	println!("loading map file..");
+	let data=read_entire_file("../tools/bhop_maps/5692113331.snfm")?;
+	let map=strafesnet_snf::read_map(data)?.into_complete_map()?;
+
+	println!("loading bot file..");
+	let data=read_entire_file("../tools/replays/535s+159764769ns.snfb")?;
+	let bot=strafesnet_snf::read_bot(data)?.read_all()?;
+
+	// create recording
+	println!("generating models..");
+	let physics_data=PhysicsData::new(&map);
+	println!("simulating...");
+	let mut physics=PhysicsState::default();
+	for ins in bot.instructions{
+		PhysicsContext::run_input_instruction(&mut physics,&physics_data,ins);
+	}
+	match physics.get_finish_time(){
+		Some(time)=>println!("finish time:{}",time),
+		None=>println!("simulation did not end in finished state"),
+	}
+
+	Ok(())
+}
+enum DeterminismResult{
+	Deterministic,
+	NonDeterministic,
+}
+struct SegmentResult{
+	determinism:DeterminismResult,
+	ticks:u64,
+	nanoseconds:u64,
+}
+fn segment_determinism(bot:strafesnet_snf::bot::Segment,physics_data:&PhysicsData)->SegmentResult{
+	// create default physics state
+	let mut physics_deterministic=PhysicsState::default();
+	// create a second physics state
+	let mut physics_filtered=PhysicsState::default();
+
+	// invent a new bot id and insert the replay
+	println!("simulating...");
+
+	let mut non_idle_count=0;
+	let instruction_count=bot.instructions.len();
+
+	let start=Instant::now();
+
+	for (i,ins) in bot.instructions.into_iter().enumerate(){
+		let state_deterministic=physics_deterministic.clone();
+		let state_filtered=physics_filtered.clone();
+		PhysicsContext::run_input_instruction(&mut physics_deterministic,physics_data,ins.clone());
+		match ins{
+			strafesnet_common::instruction::TimedInstruction{instruction:strafesnet_common::physics::Instruction::Idle,..}=>(),
+			other=>{
+				non_idle_count+=1;
+				// run
+				PhysicsContext::run_input_instruction(&mut physics_filtered,physics_data,other.clone());
+				// check if position matches
+				let b0=physics_deterministic.camera_trajectory(physics_data);
+				let b1=physics_filtered.camera_trajectory(physics_data);
+				if b0.position!=b1.position{
+					let nanoseconds=start.elapsed().as_nanos() as u64;
+					println!("desync at instruction #{}",i);
+					println!("non idle instructions completed={non_idle_count}");
+					println!("instruction #{i}={:?}",other);
+					println!("deterministic state0:\n{state_deterministic:?}");
+					println!("filtered state0:\n{state_filtered:?}");
+					println!("deterministic state1:\n{:?}",physics_deterministic);
+					println!("filtered state1:\n{:?}",physics_filtered);
+					return SegmentResult{
+						determinism:DeterminismResult::NonDeterministic,
+						ticks:1+i as u64+non_idle_count,
+						nanoseconds,
+					};
+				}
+			},
+		}
+	}
+	let nanoseconds=start.elapsed().as_nanos() as u64;
+	match physics_deterministic.get_finish_time(){
+		Some(time)=>println!("[with idle] finish time:{}",time),
+		None=>println!("[with idle] simulation did not end in finished state"),
+	}
+	match physics_filtered.get_finish_time(){
+		Some(time)=>println!("[filtered] finish time:{}",time),
+		None=>println!("[filtered] simulation did not end in finished state"),
+	}
+	SegmentResult{
+		determinism:DeterminismResult::Deterministic,
+		ticks:instruction_count as u64+non_idle_count,
+		nanoseconds,
+	}
+
+}
+type ThreadResult=Result<Option<SegmentResult>,ReplayError>;
+fn read_and_run(file_path:std::path::PathBuf,physics_data:&PhysicsData)->ThreadResult{
+	let data=read_entire_file(file_path.as_path())?;
+	let bot=strafesnet_snf::read_bot(data)?.read_all()?;
+	println!("Running {:?}",file_path.file_stem());
+	Ok(Some(segment_determinism(bot,physics_data)))
+}
+fn do_thread<'a>(s:&'a std::thread::Scope<'a,'_>,file_path:std::path::PathBuf,send:std::sync::mpsc::Sender<ThreadResult>,physics_data:&'a PhysicsData){
+	s.spawn(move ||{
+		let result=read_and_run(file_path,physics_data);
+		// send when thread is complete
+		send.send(result).unwrap();
+	});
+}
+fn get_file_path(dir_entry:std::fs::DirEntry)->Result<Option<std::path::PathBuf>,std::io::Error>{
+	Ok(dir_entry.file_type()?.is_file().then_some(
+		dir_entry.path()
+	))
+}
+fn test_determinism()->Result<(),ReplayError>{
+	let thread_limit=std::thread::available_parallelism()?.get();
+	println!("loading map file..");
+	let data=read_entire_file("../tools/bhop_maps/5692113331.snfm")?;
+	let map=strafesnet_snf::read_map(data)?.into_complete_map()?;
+
+	println!("generating models..");
+	let physics_data=PhysicsData::new(&map);
+
+	let (send,recv)=std::sync::mpsc::channel();
+
+	let mut read_dir=std::fs::read_dir("../tools/replays")?;
+
+	// promise that &physics_data will outlive the spawned threads
+	let thread_results=std::thread::scope(|s|{
+		let mut thread_results=Vec::new();
+
+		// spawn threads
+		println!("spawning up to {thread_limit} threads...");
+		let mut active_thread_count=0;
+		while active_thread_count<thread_limit{
+			if let Some(dir_entry_result)=read_dir.next(){
+				if let Some(file_path)=get_file_path(dir_entry_result?)?{
+					active_thread_count+=1;
+					do_thread(s,file_path,send.clone(),&physics_data);
+				}
+			}else{
+				break;
+			}
+		}
+
+		// spawn another thread every time a message is received from the channel
+		println!("riding parallelism wave...");
+		while let Some(dir_entry_result)=read_dir.next(){
+			if let Some(file_path)=get_file_path(dir_entry_result?)?{
+				// wait for a thread to complete
+				thread_results.push(recv.recv().unwrap());
+				do_thread(s,file_path,send.clone(),&physics_data);
+			}
+		}
+
+		// wait for remaining threads to complete
+		println!("waiting for all threads to complete...");
+		for _ in 0..active_thread_count{
+			thread_results.push(recv.recv().unwrap());
+		}
+
+		println!("done.");
+		Ok::<_,ReplayError>(thread_results)
+	})?;
+
+	// tally results
+	#[derive(Default)]
+	struct Totals{
+		deterministic:u32,
+		nondeterministic:u32,
+		invalid:u32,
+		error:u32,
+	}
+	let mut nanoseconds=0;
+	let mut ticks=0;
+	let Totals{deterministic,nondeterministic,invalid,error}=thread_results.into_iter().fold(Totals::default(),|mut totals,result|{
+		match result{
+			Ok(Some(segment_result))=>{
+				ticks+=segment_result.ticks;
+				nanoseconds+=segment_result.nanoseconds;
+				match segment_result.determinism{
+					DeterminismResult::Deterministic=>totals.deterministic+=1,
+					DeterminismResult::NonDeterministic=>totals.nondeterministic+=1,
+				}
+			},
+			Ok(None)=>totals.invalid+=1,
+			Err(_)=>totals.error+=1,
+		}
+		totals
+	});
+
+	println!("deterministic={deterministic}");
+	println!("nondeterministic={nondeterministic}");
+	println!("invalid={invalid}");
+	println!("error={error}");
+	println!("average ticks/s per core: {}",ticks*1_000_000_000/nanoseconds);
+
+	assert!(nondeterministic==0);
+	assert!(invalid==0);
+	assert!(error==0);
+
+	Ok(())
+}

integration-testing/src/test_scenes.rs

@@ -0,0 +1,113 @@
+use strafesnet_physics::physics::{InternalInstruction,PhysicsData,PhysicsState,PhysicsContext};
+use strafesnet_common::gameplay_modes::NormalizedModes;
+use strafesnet_common::gameplay_attributes::{CollisionAttributes,CollisionAttributesId};
+use strafesnet_common::integer::{vec3,mat3,Planar64Affine3,Time};
+use strafesnet_common::model::{Mesh,Model,MeshId,ModelId,RenderConfigId};
+use strafesnet_common::map::CompleteMap;
+use strafesnet_rbx_loader::primitives::{unit_cube,CubeFaceDescription};
+
+struct TestSceneBuilder{
+	meshes:Vec<Mesh>,
+	models:Vec<Model>,
+}
+impl TestSceneBuilder{
+	fn new()->Self{
+		Self{
+			meshes:Vec::new(),
+			models:Vec::new(),
+		}
+	}
+	fn push_mesh(&mut self,mesh:Mesh)->MeshId{
+		let mesh_id=self.meshes.len();
+		self.meshes.push(mesh);
+		MeshId::new(mesh_id as u32)
+	}
+	fn push_mesh_instance(&mut self,mesh:MeshId,transform:Planar64Affine3)->ModelId{
+		let model=Model{
+			mesh,
+			attributes:CollisionAttributesId::new(0),
+			color:glam::Vec4::ONE,
+			transform,
+		};
+		let model_id=self.models.len();
+		self.models.push(model);
+		ModelId::new(model_id as u32)
+	}
+	fn build(self)->PhysicsData{
+		let modes=NormalizedModes::new(Vec::new());
+		let attributes=vec![CollisionAttributes::contact_default()];
+		let meshes=self.meshes;
+		let models=self.models;
+		let textures=Vec::new();
+		let render_configs=Vec::new();
+		PhysicsData::new(&CompleteMap{
+			modes,
+			attributes,
+			meshes,
+			models,
+			textures,
+			render_configs,
+		})
+	}
+}
+
+fn test_scene()->PhysicsData{
+	let mut builder=TestSceneBuilder::new();
+	let cube_face_description=CubeFaceDescription::new(Default::default(),RenderConfigId::new(0));
+	let mesh=builder.push_mesh(unit_cube(cube_face_description));
+	// place two 5x5x5 cubes.
+	builder.push_mesh_instance(mesh,Planar64Affine3::new(
+		mat3::from_diagonal(vec3::int(5,5,5)>>1),
+		vec3::int(0,0,0)
+	));
+	builder.push_mesh_instance(mesh,Planar64Affine3::new(
+		mat3::from_diagonal(vec3::int(5,5,5)>>1),
+		vec3::int(5,-5,0)
+	));
+	builder.build()
+}
+
+#[test]
+fn simultaneous_collision(){
+	let physics_data=test_scene();
+	let body=strafesnet_physics::physics::Body::new(
+		(vec3::int(5+2,0,0)>>1)+vec3::int(1,1,0),
+		vec3::int(-1,-1,0),
+		Time::ZERO,
+	);
+	let mut physics=PhysicsState::new_with_body(body);
+	physics.style_mut().gravity=vec3::zero();
+	let mut phys_iter=PhysicsContext::iter_internal(&mut physics,&physics_data,Time::from_secs(2))
+		.filter(|ins|!matches!(ins.instruction,InternalInstruction::StrafeTick));
+	// the order that they hit does matter, but we aren't currently worrying about that.
+	// See multi-collision branch
+	assert_eq!(phys_iter.next().unwrap().time,Time::from_secs(1));
+	assert_eq!(phys_iter.next().unwrap().time,Time::from_secs(1));
+	assert!(phys_iter.next().is_none());
+	let body=physics.body();
+	assert_eq!(body.position,vec3::int(5,0,0));
+	assert_eq!(body.velocity,vec3::int(0,0,0));
+	assert_eq!(body.time,Time::from_secs(1));
+}
+#[test]
+fn bug_3(){
+	let physics_data=test_scene();
+	let body=strafesnet_physics::physics::Body::new(
+		(vec3::int(5+2,0,0)>>1)+vec3::int(1,2,0),
+		vec3::int(-1,-1,0),
+		Time::ZERO,
+	);
+	let mut physics=PhysicsState::new_with_body(body);
+	physics.style_mut().gravity=vec3::zero();
+	let mut phys_iter=PhysicsContext::iter_internal(&mut physics,&physics_data,Time::from_secs(3))
+		.filter(|ins|!matches!(ins.instruction,InternalInstruction::StrafeTick));
+	// touch side of part at 0,0,0
+	assert_eq!(phys_iter.next().unwrap().time,Time::from_secs(1));
+	// touch top of part at 5,-5,0
+	assert_eq!(phys_iter.next().unwrap().time,Time::from_secs(2));
+	assert!(phys_iter.next().is_none());
+	let body=physics.body();
+	assert_eq!(body.position,vec3::int(5+2,0,0)>>1);
+	assert_eq!(body.velocity,vec3::int(0,0,0));
+	assert_eq!(body.time,Time::from_secs(2));
+}

integration-testing/src/tests.rs

@@ -0,0 +1,78 @@
+use crate::error::ReplayError;
+use crate::util::read_entire_file;
+use strafesnet_physics::physics::{PhysicsData,PhysicsState,PhysicsContext};
+
+#[test]
+#[ignore]
+fn physics_bug_2()->Result<(),ReplayError>{
+	println!("loading map file..");
+	let data=read_entire_file("test_files/bhop_monster_jam.snfm")?;
+	let map=strafesnet_snf::read_map(data)?.into_complete_map()?;
+
+	// create recording
+	println!("generating models..");
+	let physics_data=PhysicsData::new(&map);
+	println!("simulating...");
+
+	//teleport to bug
+	// body pos = Vector { array: [Fixed { bits: 554895163352 }, Fixed { bits: 1485633089990 }, Fixed { bits: 1279601007173 }] }
+	// after the fix it's still happening, possibly for a different reason, new position to evince:
+	// body pos = Vector { array: [Fixed { bits: 555690659654 }, Fixed { bits: 1490485868773 }, Fixed { bits: 1277783839382 }] }
+	use strafesnet_common::integer::{vec3,Time};
+	let body=strafesnet_physics::physics::Body::new(
+		vec3::raw_xyz(555690659654,1490485868773,1277783839382),
+		vec3::int(0,0,0),
+		Time::ZERO,
+	);
+	let mut physics=PhysicsState::new_with_body(body);
+	// wait one second to activate the bug
+	// hit=Some(ModelId(2262))
+	PhysicsContext::run_input_instruction(&mut physics,&physics_data,strafesnet_common::instruction::TimedInstruction{
+		time:Time::from_millis(500),
+		instruction:strafesnet_common::physics::Instruction::Idle,
+	});
+
+	Ok(())
+}
+
+#[test]
+#[ignore]
+fn physics_bug_3()->Result<(),ReplayError>{
+	println!("loading map file..");
+	let data=read_entire_file("../tools/bhop_maps/5692152916.snfm")?;
+	let map=strafesnet_snf::read_map(data)?.into_complete_map()?;
+
+	// create recording
+	println!("generating models..");
+	let physics_data=PhysicsData::new(&map);
+	println!("simulating...");
+
+	//teleport to bug
+	use strafesnet_common::integer::{vec3,Time};
+	let body=strafesnet_physics::physics::Body::new(
+		// bhop_toc corner position after wall hits
+		// vec3::raw_xyz(-1401734815424,3315081280280,-2466057177493),
+		// vec3::raw_xyz(0,-96915585363,1265),
+		// vec3::raw_xyz(0,-429496729600,0),
+		// alternate room center position
+		// vec3::raw_xyz(-1129043783837,3324870327882,-2014012350212),
+		// vec3::raw_xyz(0,-96915585363,1265),
+		// vec3::raw_xyz(0,-429496729600,0),
+		// corner setup before wall hits
+		// vec3::raw_xyz(-1392580080675,3325402529458,-2444727738679),
+		// vec3::raw_xyz(-30259028820,-22950929553,-71141663007),
+		// vec3::raw_xyz(0,-429496729600,0),
+		// Actual bug 3 repro
+		vec3::raw_xyz(-2505538624455,3357963283914,557275711118),
+		vec3::raw_xyz(204188283920,-282280474198,166172785440),
+		Time::ZERO,
+	);
+	let mut physics=PhysicsState::new_with_body(body);
+	// wait one second to activate the bug
+	PhysicsContext::run_input_instruction(&mut physics,&physics_data,strafesnet_common::instruction::TimedInstruction{
+		time:Time::from_millis(500),
+		instruction:strafesnet_common::physics::Instruction::Idle,
+	});
+
+	Ok(())
+}

integration-testing/src/util.rs

@@ -0,0 +1,7 @@
+use std::io::Cursor;
+use std::path::Path;
+
+pub fn read_entire_file(path:impl AsRef<Path>)->Result<Cursor<Vec<u8>>,std::io::Error>{
+	let data=std::fs::read(path)?;
+	Ok(Cursor::new(data))
+}

lib/bsp_loader/Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 name = "strafesnet_bsp_loader"
-version = "0.2.2"
-edition = "2021"
+version = "0.5.0"
+edition = "2024"
 repository = "https://git.itzana.me/StrafesNET/strafe-project"
 license = "MIT OR Apache-2.0"
 description = "Convert Valve BSP files to StrafesNET data structures."
@@ -10,7 +10,13 @@ authors = ["Rhys Lloyd <krakow20@gmail.com>"]
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 
 [dependencies]
-glam = "0.29.0"
-strafesnet_common = { path = "../common", registry = "strafesnet" }
-vbsp = "0.6.0"
+glam.workspace = true
+strafesnet_common.workspace = true
+strafesnet_deferred_loader.workspace = true
+vbsp = "0.9.1"
+vbsp-entities-css = "0.6.0"
 vmdl = "0.2.0"
+vpk = "0.3.0"
+
+[lints]
+workspace = true

lib/bsp_loader/src/brush.rs

@@ -0,0 +1,344 @@
+use strafesnet_common::{model,integer};
+use strafesnet_common::integer::fixed_types::F192_96;
+use strafesnet_common::integer::vec3::Vector3;
+use strafesnet_common::integer::{Planar64,Planar64Vec3,Ratio};
+
+use crate::{valve_transform_normal,valve_transform_dist};
+
+#[derive(Hash,Eq,PartialEq)]
+struct Face{
+	normal:Planar64Vec3,
+	dot:Planar64,
+}
+
+#[derive(Debug)]
+struct Faces{
+	faces:Vec<Vec<Planar64Vec3>>,
+}
+
+fn solve3(c0:&Face,c1:&Face,c2:&Face)->Option<Ratio<Vector3<F192_96>,F192_96>>{
+	let n0_n1=c0.normal.cross(c1.normal);
+	let det=c2.normal.dot(n0_n1);
+	if det.abs().is_zero(){
+		return None;
+	}
+	Some((
+		c1.normal.cross(c2.normal)*c0.dot
+		+c2.normal.cross(c0.normal)*c1.dot
+		+c0.normal.cross(c1.normal)*c2.dot
+	)/det)
+}
+
+#[derive(Debug)]
+pub enum PlanesToFacesError{
+	InitFace1,
+	InitFace2,
+	InitIntersection,
+	FindNewIntersection,
+	// Narrow(strafesnet_common::integer::NarrowError),
+	EmptyFaces,
+	InfiniteLoop1,
+	InfiniteLoop2,
+}
+impl std::fmt::Display for PlanesToFacesError{
+	fn fmt(&self,f:&mut std::fmt::Formatter<'_>)->std::fmt::Result{
+		write!(f,"{self:?}")
+	}
+}
+impl core::error::Error for PlanesToFacesError{}
+
+fn planes_to_faces(face_list:std::collections::HashSet<Face>)->Result<Faces,PlanesToFacesError>{
+	let mut faces=Vec::new();
+	// for each face, determine one edge at a time until you complete the face
+	'face: for face0 in &face_list{
+		// 1. find first edge
+		// 2. follow edges around face
+
+		// === finding first edge ===
+		// 1. pick the most perpendicular set of 3 faces
+		// 2. check if any faces occlude the intersection
+		// 3. use this test to replace left and right alternating until they are not occluded
+
+		// find the most perpendicular face to face0
+		let mut face1=face_list.iter().min_by_key(|&p|{
+			face0.normal.dot(p.normal).abs()
+		}).ok_or(PlanesToFacesError::InitFace1)?;
+
+		// direction of edge formed by face0 x face1
+		let edge_dir=face0.normal.cross(face1.normal);
+
+		// find the most perpendicular face to both face0 and face1
+		let mut face2=face_list.iter().max_by_key(|&p|{
+			// find the best *oriented* face (no .abs())
+			edge_dir.dot(p.normal)
+		}).ok_or(PlanesToFacesError::InitFace2)?;
+
+		let mut detect_loop=200u8;
+
+		let mut intersection=solve3(face0,face1,face2).ok_or(PlanesToFacesError::InitIntersection)?;
+
+		// repeatedly update face1, face2 until all faces form part of the convex solid
+		'find: loop{
+			detect_loop=detect_loop.checked_sub(1).ok_or(PlanesToFacesError::InfiniteLoop1)?;
+			// test if any *other* faces occlude the intersection
+			for new_face in &face_list{
+				// new face occludes intersection point
+				if (new_face.dot.widen_128()/Planar64::ONE).lt_ratio(new_face.normal.dot(intersection.num)/intersection.den){
+					// replace one of the faces with the new face
+					// dont' try to replace face0 because we are exploring that face in particular
+					if let Some(new_intersection)=solve3(face0,new_face,face2){
+						// face1 does not occlude (or intersect) the new intersection
+						if (face1.dot.widen_128()/Planar64::ONE).gt_ratio(face1.normal.dot(new_intersection.num)/new_intersection.den){
+							face1=new_face;
+							intersection=new_intersection;
+							continue 'find;
+						}
+					}
+					if let Some(new_intersection)=solve3(face0,face1,new_face){
+						// face2 does not occlude (or intersect) the new intersection
+						if (face2.dot.widen_128()/Planar64::ONE).gt_ratio(face2.normal.dot(new_intersection.num)/new_intersection.den){
+							face2=new_face;
+							intersection=new_intersection;
+							continue 'find;
+						}
+					}
+				}
+			}
+
+			// we have found a set of faces for which the intersection is on the convex solid
+			break 'find;
+		}
+
+		// check if face0 must go, meaning it is a degenerate face and does not contribute anything to the convex solid
+		for new_face in &face_list{
+			if core::ptr::eq(face0,new_face){
+				continue;
+			}
+			if core::ptr::eq(face1,new_face){
+				continue;
+			}
+			if core::ptr::eq(face2,new_face){
+				continue;
+			}
+			// new_face occludes intersection meaning intersection is not on convex solid and face0 is degenrate
+			if (new_face.dot.widen_128()/Planar64::ONE).lt_ratio(new_face.normal.dot(intersection.num)/intersection.den){
+				// abort! reject face0 entirely
+				continue 'face;
+			}
+		}
+
+		// === follow edges around face ===
+		// Note that we chose face2 such that the 3 faces create a particular winding order.
+		// If we choose a consistent face to follow (face1, face2) it will always wind with a consistent chirality
+
+		let mut detect_loop=200u8;
+
+		// keep looping until we meet this face again
+		let face1=face1;
+		let mut face=Vec::new();
+		loop{
+			// push point onto vertices
+			// problem: this may push a vertex that does not fit in the fixed point range and is thus meaningless
+			face.push(intersection.divide().narrow_64().unwrap());
+
+			// we looped back around to face1, we're done!
+			if core::ptr::eq(face1,face2){
+				break;
+			}
+
+			// the measure
+			let edge_dir=face0.normal.cross(face2.normal);
+
+			// the dot product to beat
+			let d_intersection=edge_dir.dot(intersection.num)/intersection.den;
+
+			// find the next face moving clockwise around face0
+			let (new_face,new_intersection,_)=face_list.iter().filter_map(|new_face|{
+				// ignore faces that are part of the current edge
+				if core::ptr::eq(face0,new_face)
+				  |core::ptr::eq(face2,new_face){
+					return None;
+				}
+				let new_intersection=solve3(face0,face2,new_face)?;
+
+				// the d value must be larger
+				let d_new_intersection=edge_dir.dot(new_intersection.num)/new_intersection.den;
+				if d_new_intersection.le_ratio(d_intersection){
+					return None;
+				}
+
+				Some((new_face,new_intersection,d_new_intersection))
+			}).min_by_key(|&(_,_,d)|d).ok_or(PlanesToFacesError::FindNewIntersection)?;
+
+			face2=new_face;
+			intersection=new_intersection;
+
+			detect_loop=detect_loop.checked_sub(1).ok_or(PlanesToFacesError::InfiniteLoop2)?;
+		}
+
+		faces.push(face);
+	}
+
+	if faces.is_empty(){
+		Err(PlanesToFacesError::EmptyFaces)
+	}else{
+		Ok(Faces{
+			faces,
+		})
+	}
+}
+
+#[expect(dead_code)]
+#[derive(Debug)]
+pub enum BrushToMeshError{
+	SliceBrushSides,
+	MissingPlane,
+	InvalidFaceCount{
+		count:usize,
+	},
+	InvalidPlanes(PlanesToFacesError),
+}
+impl std::fmt::Display for BrushToMeshError{
+	fn fmt(&self,f:&mut std::fmt::Formatter<'_>)->std::fmt::Result{
+		write!(f,"{self:?}")
+	}
+}
+impl core::error::Error for BrushToMeshError{}
+
+pub fn faces_to_mesh(faces:Vec<Vec<Planar64Vec3>>)->model::Mesh{
+	// generate the mesh
+	let mut mb=model::MeshBuilder::new();
+	let color=mb.acquire_color_id(glam::Vec4::ONE);
+	let tex=mb.acquire_tex_id(glam::Vec2::ZERO);
+	// normals are ignored by physics
+	let normal=mb.acquire_normal_id(integer::vec3::zero());
+
+	let polygon_list=faces.into_iter().map(|face|{
+		face.into_iter().map(|pos|{
+			let pos=mb.acquire_pos_id(pos);
+			mb.acquire_vertex_id(model::IndexedVertex{
+				pos,
+				tex,
+				normal,
+				color,
+			})
+		}).collect()
+	}).collect();
+
+	let polygon_groups=vec![model::PolygonGroup::PolygonList(model::PolygonList::new(polygon_list))];
+	let physics_groups=vec![model::IndexedPhysicsGroup{
+		groups:vec![model::PolygonGroupId::new(0)],
+	}];
+	let graphics_groups=vec![];
+
+	mb.build(polygon_groups,graphics_groups,physics_groups)
+}
+
+pub fn brush_to_mesh(bsp:&vbsp::Bsp,brush:&vbsp::Brush)->Result<model::Mesh,BrushToMeshError>{
+	let brush_start_idx=brush.brush_side as usize;
+	let sides_range=brush_start_idx..brush_start_idx+brush.num_brush_sides as usize;
+	let sides=bsp.brush_sides.get(sides_range).ok_or(BrushToMeshError::SliceBrushSides)?;
+	let face_list=sides.iter().map(|side|{
+		// The so-called tumor brushes have TRIGGER bit set
+		// but also ignore visleaf hint and skip sides
+		const TUMOR:vbsp::TextureFlags=vbsp::TextureFlags::HINT.union(vbsp::TextureFlags::SKIP).union(vbsp::TextureFlags::TRIGGER);
+		if let Some(texture_info)=bsp.texture_info(side.texture_info as usize){
+			if texture_info.flags.intersects(TUMOR){
+				return None;
+			}
+		}
+		let plane=bsp.plane(side.plane as usize)?;
+		Some(Face{
+			normal:valve_transform_normal(plane.normal.into()),
+			dot:valve_transform_dist(plane.dist.into()),
+		})
+	}).collect::<Option<std::collections::HashSet<_>>>().ok_or(BrushToMeshError::MissingPlane)?;
+
+	if face_list.len()<4{
+		return Err(BrushToMeshError::InvalidFaceCount{count:face_list.len()});
+	}
+
+	let faces=planes_to_faces(face_list).map_err(BrushToMeshError::InvalidPlanes)?;
+
+	let mesh=faces_to_mesh(faces.faces);
+
+	Ok(mesh)
+}
+
+pub fn unit_cube()->model::Mesh{
+	let face_list=[
+		Face{normal:integer::vec3::X,dot:Planar64::ONE},
+		Face{normal:integer::vec3::Y,dot:Planar64::ONE},
+		Face{normal:integer::vec3::Z,dot:Planar64::ONE},
+		Face{normal:integer::vec3::NEG_X,dot:Planar64::ONE},
+		Face{normal:integer::vec3::NEG_Y,dot:Planar64::ONE},
+		Face{normal:integer::vec3::NEG_Z,dot:Planar64::ONE},
+	].into_iter().collect();
+	let faces=planes_to_faces(face_list).unwrap();
+	let mesh=faces_to_mesh(faces.faces);
+	mesh
+}
+
+#[cfg(test)]
+mod test{
+	use super::*;
+	#[test]
+	fn test_cube(){
+		let face_list=[
+			Face{normal:integer::vec3::X,dot:Planar64::ONE},
+			Face{normal:integer::vec3::Y,dot:Planar64::ONE},
+			Face{normal:integer::vec3::Z,dot:Planar64::ONE},
+			Face{normal:integer::vec3::NEG_X,dot:Planar64::ONE},
+			Face{normal:integer::vec3::NEG_Y,dot:Planar64::ONE},
+			Face{normal:integer::vec3::NEG_Z,dot:Planar64::ONE},
+		].into_iter().collect();
+		let faces=planes_to_faces(face_list).unwrap();
+		assert_eq!(faces.faces.len(),6);
+		dbg!(faces);
+	}
+	#[test]
+	fn test_cube_with_degernate_face(){
+		let face_list=[
+			Face{normal:integer::vec3::X,dot:Planar64::ONE},
+			Face{normal:integer::vec3::Y,dot:Planar64::ONE},
+			Face{normal:integer::vec3::Z,dot:Planar64::ONE},
+			Face{normal:integer::vec3::NEG_X,dot:Planar64::ONE},
+			Face{normal:integer::vec3::NEG_Y,dot:Planar64::ONE},
+			Face{normal:integer::vec3::NEG_Z,dot:Planar64::ONE},
+			Face{normal:integer::vec3::NEG_Z,dot:Planar64::EPSILON},
+		].into_iter().collect();
+		let faces=planes_to_faces(face_list).unwrap();
+		assert_eq!(faces.faces.len(),6);
+		dbg!(faces);
+	}
+	#[test]
+	fn test_cube_with_degernate_face2(){
+		let face_list=[
+			Face{normal:integer::vec3::X,dot:Planar64::ONE},
+			Face{normal:integer::vec3::Y,dot:Planar64::ONE},
+			Face{normal:integer::vec3::Z,dot:Planar64::ONE},
+			Face{normal:integer::vec3::NEG_X,dot:Planar64::ONE},
+			Face{normal:integer::vec3::NEG_Y,dot:Planar64::ONE},
+			Face{normal:integer::vec3::NEG_Z,dot:Planar64::ONE},
+			Face{normal:integer::vec3::NEG_X+integer::vec3::NEG_Z,dot:-Planar64::EPSILON},
+		].into_iter().collect();
+		let faces=planes_to_faces(face_list).unwrap();
+		assert_eq!(faces.faces.len(),5);
+		dbg!(faces);
+	}
+	#[test]
+	fn test_cube_with_degernate_face3(){
+		let face_list=[
+			Face{normal:integer::vec3::X,dot:Planar64::ONE},
+			Face{normal:integer::vec3::Y,dot:Planar64::ONE},
+			Face{normal:integer::vec3::Z,dot:Planar64::ONE},
+			Face{normal:integer::vec3::NEG_X,dot:Planar64::ONE},
+			Face{normal:integer::vec3::NEG_Y,dot:Planar64::ONE},
+			Face{normal:integer::vec3::NEG_Z,dot:Planar64::ONE},
+			Face{normal:integer::vec3::NEG_X+integer::vec3::NEG_Z,dot:Planar64::EPSILON},
+		].into_iter().collect();
+		let faces=planes_to_faces(face_list).unwrap();
+		assert_eq!(faces.faces.len(),7);
+		dbg!(faces);
+	}
+}

lib/bsp_loader/src/bsp.rs

@@ -1,43 +1,142 @@
-use strafesnet_common::{map,model,integer,gameplay_attributes};
+use std::borrow::Cow;
 
-const VALVE_SCALE:f32=1.0/16.0;
-fn valve_transform([x,y,z]:[f32;3])->integer::Planar64Vec3{
-	integer::vec3::try_from_f32_array([x*VALVE_SCALE,z*VALVE_SCALE,-y*VALVE_SCALE]).unwrap()
+use vbsp_entities_css::Entity;
+
+use strafesnet_common::{map,model,integer,gameplay_attributes as attr};
+use strafesnet_deferred_loader::deferred_loader::{MeshDeferredLoader,RenderConfigDeferredLoader};
+use strafesnet_deferred_loader::mesh::Meshes;
+use strafesnet_deferred_loader::texture::{RenderConfigs,Texture};
+use strafesnet_common::gameplay_modes::{NormalizedMode,NormalizedModes,Mode,Stage};
+
+use crate::valve_transform;
+
+fn ingest_vertex(
+	mb:&mut model::MeshBuilder,
+	world_position:vbsp::Vector,
+	texture_transform_u:glam::Vec4,
+	texture_transform_v:glam::Vec4,
+	normal:model::NormalId,
+	color:model::ColorId,
+)->model::VertexId{
+	//world_model.origin seems to always be 0,0,0
+	let vertex_xyz=world_position.into();
+	let pos=mb.acquire_pos_id(valve_transform(vertex_xyz));
+
+	//calculate texture coordinates
+	let pos_4d=glam::Vec3::from_array(vertex_xyz).extend(1.0);
+	let tex=glam::vec2(texture_transform_u.dot(pos_4d),texture_transform_v.dot(pos_4d));
+	let tex=mb.acquire_tex_id(tex);
+
+	mb.acquire_vertex_id(model::IndexedVertex{
+		pos,
+		tex,
+		normal,
+		color,
+	})
 }
-pub fn convert_bsp<AcquireRenderConfigId,AcquireMeshId>(
-	bsp:&vbsp::Bsp,
-	mut acquire_render_config_id:AcquireRenderConfigId,
-	mut acquire_mesh_id:AcquireMeshId
-)->PartialMap1
-where
-	AcquireRenderConfigId:FnMut(Option<&str>)->model::RenderConfigId,
-	AcquireMeshId:FnMut(&str)->model::MeshId,
-{
-	//figure out real attributes later
-	let mut unique_attributes=Vec::new();
-	unique_attributes.push(gameplay_attributes::CollisionAttributes::Decoration);
-	const TEMP_TOUCH_ME_ATTRIBUTE:gameplay_attributes::CollisionAttributesId=gameplay_attributes::CollisionAttributesId::new(0);
 
-	let mut prop_mesh_count=0;
+fn add_brush<'a>(
+	mesh_deferred_loader:&mut MeshDeferredLoader<&'a str>,
+	world_models:&mut Vec<model::Model>,
+	prop_models:&mut Vec<model::Model>,
+	model:&'a str,
+	origin:vbsp::Vector,
+	rendercolor:vbsp::Color,
+	attributes:attr::CollisionAttributesId,
+){
+	let transform=integer::Planar64Affine3::from_translation(
+		valve_transform(origin.into())
+	);
+	let color=(glam::Vec3::from_array([
+		rendercolor.r as f32,
+		rendercolor.g as f32,
+		rendercolor.b as f32
+	])/255.0).extend(1.0);
+
+	match model.chars().next(){
+		// The first character of brush.model is '*'
+		Some('*')=>match model[1..].parse(){
+			Ok(mesh_id)=>{
+				let mesh=model::MeshId::new(mesh_id);
+				world_models.push(
+					model::Model{mesh,attributes,transform,color}
+				);
+			},
+			Err(e)=>{
+				println!("Brush model int parse error: {e} model={model}");
+				return;
+			},
+		},
+		_=>{
+			let mesh=mesh_deferred_loader.acquire_mesh_id(model);
+			prop_models.push(
+				model::Model{mesh,attributes,transform,color}
+			);
+		}
+	}
+}
+
+pub fn convert<'a>(
+	bsp:&'a crate::Bsp,
+	render_config_deferred_loader:&mut RenderConfigDeferredLoader<Cow<'a,str>>,
+	mesh_deferred_loader:&mut MeshDeferredLoader<&'a str>,
+)->PartialMap1{
+	let bsp=bsp.as_ref();
+	//figure out real attributes later
+	let unique_attributes=vec![
+		attr::CollisionAttributes::Decoration,
+		attr::CollisionAttributes::contact_default(),
+		attr::CollisionAttributes::intersect_default(),
+		// ladder
+		attr::CollisionAttributes::Contact(
+			attr::ContactAttributes{
+				contacting:attr::ContactingAttributes{
+					contact_behaviour:Some(attr::ContactingBehaviour::Ladder(
+						attr::ContactingLadder{sticky:true}
+					))
+				},
+				general:attr::GeneralAttributes::default(),
+			}
+		),
+		// water
+		attr::CollisionAttributes::Intersect(
+			attr::IntersectAttributes{
+				intersecting:attr::IntersectingAttributes{
+					water:Some(attr::IntersectingWater{
+						viscosity:integer::Planar64::ONE,
+						density:integer::Planar64::ONE,
+						velocity:integer::vec3::zero(),
+					}),
+				},
+				general:attr::GeneralAttributes::default(),
+			}
+		),
+	];
+	const ATTRIBUTE_DECORATION:attr::CollisionAttributesId=attr::CollisionAttributesId::new(0);
+	const ATTRIBUTE_CONTACT_DEFAULT:attr::CollisionAttributesId=attr::CollisionAttributesId::new(1);
+	const ATTRIBUTE_INTERSECT_DEFAULT:attr::CollisionAttributesId=attr::CollisionAttributesId::new(2);
+	const ATTRIBUTE_LADDER_DEFAULT:attr::CollisionAttributesId=attr::CollisionAttributesId::new(3);
+	const ATTRIBUTE_WATER_DEFAULT:attr::CollisionAttributesId=attr::CollisionAttributesId::new(4);
+
 	//declare all prop models to Loader
-	let prop_models=bsp.static_props().map(|prop|{
+	let mut prop_models=bsp.static_props().map(|prop|{
+		const DEG_TO_RAD:f32=std::f32::consts::TAU/360.0;
 		//get or create mesh_id
-		let mesh_id=acquire_mesh_id(prop.model());
-		//not the most failsafe code but this is just for the map tool lmao
-		if prop_mesh_count==mesh_id.get(){
-			prop_mesh_count+=1;
-		};
-		let placement=prop.as_prop_placement();
+		let mesh_id=mesh_deferred_loader.acquire_mesh_id(prop.model());
 		model::Model{
 			mesh:mesh_id,
-			attributes:TEMP_TOUCH_ME_ATTRIBUTE,
+			attributes:ATTRIBUTE_DECORATION,
 			transform:integer::Planar64Affine3::new(
-				integer::mat3::try_from_f32_array_2d((
-					glam::Mat3A::from_diagonal(glam::Vec3::splat(placement.scale))
+				integer::mat3::try_from_f32_array_2d(
 					//TODO: figure this out
-					*glam::Mat3A::from_quat(glam::Quat::from_array(placement.rotation.into()))
-				).to_cols_array_2d()).unwrap(),
-				valve_transform(placement.origin.into()),
+					glam::Mat3A::from_euler(
+						glam::EulerRot::XYZ,
+						prop.angles.pitch*DEG_TO_RAD,
+						prop.angles.yaw*DEG_TO_RAD,
+						prop.angles.roll*DEG_TO_RAD
+					).to_cols_array_2d()
+				).unwrap(),
+				valve_transform(prop.origin.into()),
 			),
 			color:glam::Vec4::ONE,
 		}
@@ -47,14 +146,12 @@ where
 
 	//the generated MeshIds in here will collide with the Loader Mesh Ids
 	//but I can't think of a good workaround other than just remapping one later.
-	let world_meshes:Vec<model::Mesh>=bsp.models().map(|world_model|{
-		//non-deduplicated
-		let mut spam_pos=Vec::new();
-		let mut spam_tex=Vec::new();
-		let mut spam_normal=Vec::new();
-		let mut spam_vertices=Vec::new();
+	let mut world_meshes:Vec<model::Mesh>=bsp.models().map(|world_model|{
+		let mut mb=model::MeshBuilder::new();
+
+		let color=mb.acquire_color_id(glam::Vec4::ONE);
 		let mut graphics_groups=Vec::new();
-		let mut physics_group=model::IndexedPhysicsGroup::default();
+		let mut render_id_to_graphics_group_id=std::collections::HashMap::new();
 		let polygon_groups=world_model.faces().enumerate().map(|(polygon_group_id,face)|{
 			let polygon_group_id=model::PolygonGroupId::new(polygon_group_id as u32);
 			let face_texture=face.texture();
@@ -63,139 +160,200 @@ where
 			let texture_transform_u=glam::Vec4::from_array(face_texture.texture_transforms_u)/(face_texture_data.width as f32);
 			let texture_transform_v=glam::Vec4::from_array(face_texture.texture_transforms_v)/(face_texture_data.height as f32);
 
-			//this automatically figures out what the texture is trying to do and creates
-			//a render config for it, and then returns the id to that render config
-			let render_id=acquire_render_config_id(Some(face_texture_data.name()));
-
 			//normal
-			let normal=face.normal();
-			let normal_idx=spam_normal.len() as u32;
-			spam_normal.push(valve_transform(normal.into()));
-			let mut polygon_iter=face.vertex_positions().map(|vertex_position|{
-				//world_model.origin seems to always be 0,0,0
-				let vertex_xyz=(world_model.origin+vertex_position).into();
-				let pos_idx=spam_pos.len();
-				spam_pos.push(valve_transform(vertex_xyz));
-
-				//calculate texture coordinates
-				let pos=glam::Vec3::from_array(vertex_xyz).extend(1.0);
-				let tex=glam::vec2(texture_transform_u.dot(pos),texture_transform_v.dot(pos));
-				let tex_idx=spam_tex.len() as u32;
-				spam_tex.push(tex);
-
-				let vertex_id=model::VertexId::new(spam_vertices.len() as u32);
-				spam_vertices.push(model::IndexedVertex{
-					pos:model::PositionId::new(pos_idx as u32),
-					tex:model::TextureCoordinateId::new(tex_idx as u32),
-					normal:model::NormalId::new(normal_idx),
-					color:model::ColorId::new(0),
-				});
-				vertex_id
-			});
+			let normal=mb.acquire_normal_id(valve_transform(face.normal().into()));
+			let mut polygon_iter=face.vertex_positions().map(|vertex_position|
+				world_model.origin+vertex_position
+			);
 			let polygon_list=std::iter::from_fn(move||{
 				match (polygon_iter.next(),polygon_iter.next(),polygon_iter.next()){
-					(Some(v1),Some(v2),Some(v3))=>Some(vec![v1,v2,v3]),
+					(Some(v1),Some(v2),Some(v3))=>Some([v1,v2,v3]),
 					//ignore extra vertices, not sure what to do in this case, failing the whole conversion could be appropriate
 					_=>None,
 				}
+			}).map(|triplet|{
+				triplet.map(|world_position|
+					ingest_vertex(&mut mb,world_position,texture_transform_u,texture_transform_v,normal,color)
+				).to_vec()
 			}).collect();
 			if face.is_visible(){
-				//TODO: deduplicate graphics groups by render id
-				graphics_groups.push(model::IndexedGraphicsGroup{
-					render:render_id,
-					groups:vec![polygon_group_id],
-				})
+				//this automatically figures out what the texture is trying to do and creates
+				//a render config for it, and then returns the id to that render config
+				let render_id=render_config_deferred_loader.acquire_render_config_id(Some(Cow::Borrowed(face_texture_data.name())));
+				//deduplicate graphics groups by render id
+				let graphics_group_id=*render_id_to_graphics_group_id.entry(render_id).or_insert_with(||{
+					let graphics_group_id=graphics_groups.len();
+					graphics_groups.push(model::IndexedGraphicsGroup{
+						render:render_id,
+						groups:vec![],
+					});
+					graphics_group_id
+				});
+				graphics_groups[graphics_group_id].groups.push(polygon_group_id);
 			}
-			physics_group.groups.push(polygon_group_id);
 			model::PolygonGroup::PolygonList(model::PolygonList::new(polygon_list))
 		}).collect();
-		model::Mesh{
-			unique_pos:spam_pos,
-			unique_tex:spam_tex,
-			unique_normal:spam_normal,
-			unique_color:vec![glam::Vec4::ONE],
-			unique_vertices:spam_vertices,
-			polygon_groups,
-			graphics_groups,
-			physics_groups:vec![physics_group],
-		}
+
+		mb.build(polygon_groups,graphics_groups,vec![])
 	}).collect();
 
-	let world_models:Vec<model::Model>=
-	//one instance of the main world mesh
-	std::iter::once((
-		//world_model
-		model::MeshId::new(0),
-		//model_origin
-		vbsp::Vector::from([0.0,0.0,0.0]),
-		//model_color
-		vbsp::Color{r:255,g:255,b:255},
-	)).chain(
-		//entities sprinkle instances of the other meshes around
-		bsp.entities.iter()
-		.flat_map(|ent|ent.parse())//ignore entity parsing errors
-		.filter_map(|ent|match ent{
-			vbsp::Entity::Brush(brush)=>Some(brush),
-			vbsp::Entity::BrushIllusionary(brush)=>Some(brush),
-			vbsp::Entity::BrushWall(brush)=>Some(brush),
-			vbsp::Entity::BrushWallToggle(brush)=>Some(brush),
-			_=>None,
-		}).flat_map(|brush|
-			//The first character of brush.model is '*'
-			brush.model[1..].parse().map(|mesh_id|//ignore parse int errors
-				(model::MeshId::new(mesh_id),brush.origin,brush.color)
-			)
-		)
-	).map(|(mesh_id,model_origin,vbsp::Color{r,g,b})|{
-		model::Model{
-			mesh:mesh_id,
-			attributes:TEMP_TOUCH_ME_ATTRIBUTE,
-			transform:integer::Planar64Affine3::new(
-				integer::mat3::identity(),
-				valve_transform(model_origin.into())
-			),
-			color:(glam::Vec3::from_array([r as f32,g as f32,b as f32])/255.0).extend(1.0),
+	let mut found_spawn=None;
+
+	let mut world_models=Vec::new();
+
+	// the one and only world model 0
+	world_models.push(model::Model{
+		mesh:model::MeshId::new(0),
+		attributes:ATTRIBUTE_DECORATION,
+		transform:integer::Planar64Affine3::IDENTITY,
+		color:glam::Vec4::W,
+	});
+
+	const WHITE:vbsp::Color=vbsp::Color{r:255,g:255,b:255};
+	for raw_ent in &bsp.entities{
+		match raw_ent.parse(){
+			Ok(Entity::Cycler(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,brush.rendercolor,ATTRIBUTE_DECORATION),
+			Ok(Entity::EnvSprite(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,brush.rendercolor,ATTRIBUTE_DECORATION),
+			Ok(Entity::FuncBreakable(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,brush.rendercolor,ATTRIBUTE_DECORATION),
+			Ok(Entity::FuncBrush(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,brush.rendercolor,ATTRIBUTE_DECORATION),
+			Ok(Entity::FuncButton(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,brush.rendercolor,ATTRIBUTE_DECORATION),
+			Ok(Entity::FuncDoor(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,brush.rendercolor,ATTRIBUTE_DECORATION),
+			Ok(Entity::FuncDoorRotating(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,brush.rendercolor,ATTRIBUTE_DECORATION),
+			Ok(Entity::FuncIllusionary(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,brush.rendercolor,ATTRIBUTE_DECORATION),
+			Ok(Entity::FuncMonitor(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,brush.rendercolor,ATTRIBUTE_DECORATION),
+			Ok(Entity::FuncMovelinear(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,brush.rendercolor,ATTRIBUTE_DECORATION),
+			Ok(Entity::FuncPhysbox(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,brush.rendercolor,ATTRIBUTE_DECORATION),
+			Ok(Entity::FuncPhysboxMultiplayer(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,brush.rendercolor,ATTRIBUTE_DECORATION),
+			Ok(Entity::FuncRotButton(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,WHITE,ATTRIBUTE_DECORATION),
+			Ok(Entity::FuncRotating(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,brush.rendercolor,ATTRIBUTE_DECORATION),
+			Ok(Entity::FuncTracktrain(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,brush.rendercolor,ATTRIBUTE_DECORATION),
+			Ok(Entity::FuncTrain(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,brush.rendercolor,ATTRIBUTE_DECORATION),
+			Ok(Entity::FuncWall(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin.unwrap_or_default(),brush.rendercolor,ATTRIBUTE_DECORATION),
+			Ok(Entity::FuncWallToggle(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin.unwrap_or_default(),brush.rendercolor,ATTRIBUTE_DECORATION),
+			Ok(Entity::FuncWaterAnalog(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,brush.rendercolor.unwrap_or(WHITE),ATTRIBUTE_DECORATION),
+			Ok(Entity::PropDoorRotating(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,WHITE,ATTRIBUTE_DECORATION),
+			Ok(Entity::PropDynamic(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,WHITE,ATTRIBUTE_DECORATION),
+			Ok(Entity::PropDynamicOverride(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,WHITE,ATTRIBUTE_DECORATION),
+			Ok(Entity::PropPhysics(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,WHITE,ATTRIBUTE_DECORATION),
+			Ok(Entity::PropPhysicsMultiplayer(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,WHITE,ATTRIBUTE_DECORATION),
+			Ok(Entity::PropPhysicsOverride(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,WHITE,ATTRIBUTE_DECORATION),
+			Ok(Entity::PropRagdoll(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,WHITE,ATTRIBUTE_DECORATION),
+			Ok(Entity::TriggerGravity(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,WHITE,ATTRIBUTE_DECORATION),
+			Ok(Entity::TriggerHurt(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,WHITE,ATTRIBUTE_DECORATION),
+			Ok(Entity::TriggerLook(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,WHITE,ATTRIBUTE_DECORATION),
+			Ok(Entity::TriggerMultiple(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model.unwrap_or_default(),brush.origin,WHITE,ATTRIBUTE_DECORATION),
+			Ok(Entity::TriggerOnce(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,WHITE,ATTRIBUTE_DECORATION),
+			Ok(Entity::TriggerProximity(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,WHITE,ATTRIBUTE_DECORATION),
+			Ok(Entity::TriggerPush(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,WHITE,ATTRIBUTE_DECORATION),
+			Ok(Entity::TriggerSoundscape(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,WHITE,ATTRIBUTE_DECORATION),
+			Ok(Entity::TriggerTeleport(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model.unwrap_or_default(),brush.origin,WHITE,ATTRIBUTE_DECORATION),
+			Ok(Entity::TriggerVphysicsMotion(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,WHITE,ATTRIBUTE_DECORATION),
+			Ok(Entity::TriggerWind(brush))=>add_brush(mesh_deferred_loader,&mut world_models,&mut prop_models,brush.model,brush.origin,WHITE,ATTRIBUTE_DECORATION),
+			Ok(Entity::InfoPlayerCounterterrorist(spawn))=>{
+				found_spawn=Some(valve_transform(spawn.origin.into()));
+			},
+			Ok(Entity::InfoPlayerTerrorist(spawn))=>{
+				found_spawn=Some(valve_transform(spawn.origin.into()));
+			},
+			Err(e)=>{
+				println!("Bsp Entity parse error: {e}");
+			},
+			_=>(),
 		}
-	}).collect();
+	}
+
+	// physics models
+	for brush in &bsp.brushes{
+		const RELEVANT:vbsp::BrushFlags=
+			vbsp::BrushFlags::SOLID
+			.union(vbsp::BrushFlags::PLAYERCLIP)
+			.union(vbsp::BrushFlags::WATER)
+			.union(vbsp::BrushFlags::MOVEABLE)
+			.union(vbsp::BrushFlags::LADDER);
+		if !brush.flags.intersects(RELEVANT){
+			continue;
+		}
+		let is_ladder=brush.flags.contains(vbsp::BrushFlags::LADDER);
+		let is_water=brush.flags.contains(vbsp::BrushFlags::WATER);
+		let attributes=match (is_ladder,is_water){
+			(true,false)=>ATTRIBUTE_LADDER_DEFAULT,
+			(false,true)=>ATTRIBUTE_WATER_DEFAULT,
+			(false,false)=>ATTRIBUTE_CONTACT_DEFAULT,
+			(true,true)=>{
+				// water ladder? wtf
+				println!("brush is a water ladder o_o defaulting to ladder");
+				ATTRIBUTE_LADDER_DEFAULT
+			}
+		};
+		let mesh_result=crate::brush::brush_to_mesh(bsp,brush);
+		match mesh_result{
+			Ok(mesh)=>{
+				let mesh_id=model::MeshId::new(world_meshes.len() as u32);
+				world_meshes.push(mesh);
+				world_models.push(model::Model{
+					mesh:mesh_id,
+					attributes,
+					transform:integer::Planar64Affine3::new(
+						integer::mat3::identity(),
+						integer::vec3::zero(),
+					),
+					color:glam::Vec4::ONE,
+				});
+			},
+			Err(e)=>println!("Brush mesh error: {e}"),
+		}
+	}
+
+	let mut modes_list=Vec::new();
+	if let Some(spawn_point)=found_spawn{
+		// create a new mesh
+		let mesh_id=model::MeshId::new(world_meshes.len() as u32);
+		world_meshes.push(crate::brush::unit_cube());
+		// create a new model
+		let model_id=model::ModelId::new(world_models.len() as u32);
+		world_models.push(model::Model{
+			mesh:mesh_id,
+			attributes:ATTRIBUTE_INTERSECT_DEFAULT,
+			transform:integer::Planar64Affine3::from_translation(spawn_point),
+			color:glam::Vec4::W,
+		});
+
+		let first_stage=Stage::empty(model_id);
+		let main_mode=Mode::new(
+			strafesnet_common::gameplay_style::StyleModifiers::source_bhop(),
+			model_id,
+			std::collections::HashMap::new(),
+			vec![first_stage],
+			std::collections::HashMap::new(),
+		);
+		modes_list.push(NormalizedMode::new(main_mode));
+	}
 
 	PartialMap1{
 		attributes:unique_attributes,
 		world_meshes,
 		prop_models,
 		world_models,
-		modes:strafesnet_common::gameplay_modes::Modes::new(Vec::new()),
+		modes:NormalizedModes::new(modes_list),
 	}
 }
 
 //partially constructed map types
 pub struct PartialMap1{
-	attributes:Vec<strafesnet_common::gameplay_attributes::CollisionAttributes>,
+	attributes:Vec<attr::CollisionAttributes>,
 	prop_models:Vec<model::Model>,
 	world_meshes:Vec<model::Mesh>,
 	world_models:Vec<model::Model>,
-	modes:strafesnet_common::gameplay_modes::Modes,
+	modes:NormalizedModes,
 }
 impl PartialMap1{
-	pub fn add_prop_meshes<AcquireRenderConfigId>(
+	pub fn add_prop_meshes(
 		self,
-		prop_meshes:impl IntoIterator<Item=(model::MeshId,crate::data::ModelData)>,
-		mut acquire_render_config_id:AcquireRenderConfigId,
-	)->PartialMap2
-	where
-		AcquireRenderConfigId:FnMut(Option<&str>)->model::RenderConfigId,
-	{
+		prop_meshes:Meshes<model::Mesh>,
+	)->PartialMap2{
 		PartialMap2{
 			attributes:self.attributes,
-			prop_meshes:prop_meshes.into_iter().filter_map(|(mesh_id,model_data)|
-				//this will generate new render ids and texture ids
-				match convert_mesh(model_data,&mut acquire_render_config_id){
-					Ok(mesh)=>Some((mesh_id,mesh)),
-					Err(e)=>{
-						println!("error converting mesh: {e}");
-						None
-					}
-				}
-			).collect(),
+			prop_meshes:prop_meshes.consume().collect(),
 			prop_models:self.prop_models,
 			world_meshes:self.world_meshes,
 			world_models:self.world_models,
@@ -204,18 +362,17 @@ impl PartialMap1{
 	}
 }
 pub struct PartialMap2{
-	attributes:Vec<strafesnet_common::gameplay_attributes::CollisionAttributes>,
+	attributes:Vec<attr::CollisionAttributes>,
 	prop_meshes:Vec<(model::MeshId,model::Mesh)>,
 	prop_models:Vec<model::Model>,
 	world_meshes:Vec<model::Mesh>,
 	world_models:Vec<model::Model>,
-	modes:strafesnet_common::gameplay_modes::Modes,
+	modes:NormalizedModes,
 }
 impl PartialMap2{
 	pub fn add_render_configs_and_textures(
 		mut self,
-		render_configs:impl IntoIterator<Item=(model::RenderConfigId,model::RenderConfig)>,
-		textures:impl IntoIterator<Item=(model::TextureId,Vec<u8>)>,
+		render_configs:RenderConfigs,
 	)->map::CompleteMap{
 		//merge mesh and model lists, flatten and remap all ids
 		let mesh_id_offset=self.world_meshes.len();
@@ -234,13 +391,14 @@ impl PartialMap2{
 			})
 		));
 		//let mut models=Vec::new();
+		let (textures,render_configs)=render_configs.consume();
 		let (textures,texture_id_map):(Vec<Vec<u8>>,std::collections::HashMap<model::TextureId,model::TextureId>)
 		=textures.into_iter()
 		//.filter_map(f) cull unused textures
-		.enumerate().map(|(new_texture_id,(old_texture_id,texture))|{
+		.enumerate().map(|(new_texture_id,(old_texture_id,Texture::ImageDDS(texture)))|{
 			(texture,(old_texture_id,model::TextureId::new(new_texture_id as u32)))
 		}).unzip();
-		let render_configs=render_configs.into_iter().map(|(render_config_id,mut render_config)|{
+		let render_configs=render_configs.into_iter().map(|(_render_config_id,mut render_config)|{
 			//this may generate duplicate no-texture render configs but idc
 			render_config.texture=render_config.texture.and_then(|texture_id|
 				texture_id_map.get(&texture_id).copied()
@@ -257,77 +415,3 @@ impl PartialMap2{
 		}
 	}
 }
-
-fn convert_mesh<AcquireRenderConfigId>(
-	model_data:crate::data::ModelData,
-	acquire_render_config_id:&mut AcquireRenderConfigId,
-)->Result<model::Mesh,vmdl::ModelError>
-where
-	AcquireRenderConfigId:FnMut(Option<&str>)->model::RenderConfigId,
-{
-	let model=model_data.read_model()?;
-	let texture_paths=model.texture_directories();
-	if texture_paths.len()!=1{
-		println!("WARNING: multiple texture paths");
-	}
-	let skin=model.skin_tables().nth(0).unwrap();
-
-	let mut spam_pos=Vec::with_capacity(model.vertices().len());
-	let mut spam_normal=Vec::with_capacity(model.vertices().len());
-	let mut spam_tex=Vec::with_capacity(model.vertices().len());
-	let mut spam_vertices=Vec::with_capacity(model.vertices().len());
-	for (i,vertex) in model.vertices().iter().enumerate(){
-		spam_pos.push(valve_transform(vertex.position.into()));
-		spam_normal.push(valve_transform(vertex.normal.into()));
-		spam_tex.push(glam::Vec2::from_array(vertex.texture_coordinates));
-		spam_vertices.push(model::IndexedVertex{
-			pos:model::PositionId::new(i as u32),
-			tex:model::TextureCoordinateId::new(i as u32),
-			normal:model::NormalId::new(i as u32),
-			color:model::ColorId::new(0),
-		});
-	}
-	let mut graphics_groups=Vec::new();
-	let mut physics_groups=Vec::new();
-	let polygon_groups=model.meshes().enumerate().map(|(polygon_group_id,mesh)|{
-		let polygon_group_id=model::PolygonGroupId::new(polygon_group_id as u32);
-
-		let render_id=if let (Some(texture_path),Some(texture_name))=(texture_paths.get(0),skin.texture(mesh.material_index())){
-			let mut path=std::path::PathBuf::from(texture_path.as_str());
-			path.push(texture_name);
-			acquire_render_config_id(path.as_os_str().to_str())
-		}else{
-			acquire_render_config_id(None)
-		};
-
-		graphics_groups.push(model::IndexedGraphicsGroup{
-			render:render_id,
-			groups:vec![polygon_group_id],
-		});
-		physics_groups.push(model::IndexedPhysicsGroup{
-			groups:vec![polygon_group_id],
-		});
-		model::PolygonGroup::PolygonList(model::PolygonList::new(
-			//looking at the code, it would seem that the strips are pre-deindexed into triangle lists when calling this function
-			mesh.vertex_strip_indices().flat_map(|mut strip|
-				std::iter::from_fn(move||{
-					match (strip.next(),strip.next(),strip.next()){
-						(Some(v1),Some(v2),Some(v3))=>Some([v1,v2,v3].map(|vertex_id|model::VertexId::new(vertex_id as u32)).to_vec()),
-						//ignore extra vertices, not sure what to do in this case, failing the whole conversion could be appropriate
-						_=>None,
-					}
-				})
-			).collect()
-		))
-	}).collect();
-	Ok(model::Mesh{
-		unique_pos:spam_pos,
-		unique_normal:spam_normal,
-		unique_tex:spam_tex,
-		unique_color:vec![glam::Vec4::ONE],
-		unique_vertices:spam_vertices,
-		polygon_groups,
-		graphics_groups,
-		physics_groups,
-	})
-}

lib/bsp_loader/src/data.rs

@@ -1,60 +0,0 @@
-pub struct Bsp(vbsp::Bsp);
-impl Bsp{
-	pub const fn new(value:vbsp::Bsp)->Self{
-		Self(value)
-	}
-}
-impl AsRef<vbsp::Bsp> for Bsp{
-	fn as_ref(&self)->&vbsp::Bsp{
-		&self.0
-	}
-}
-
-pub struct MdlData(Vec<u8>);
-impl MdlData{
-	pub const fn new(value:Vec<u8>)->Self{
-		Self(value)
-	}
-}
-impl AsRef<[u8]> for MdlData{
-	fn as_ref(&self)->&[u8]{
-		self.0.as_ref()
-	}
-}
-pub struct VtxData(Vec<u8>);
-impl VtxData{
-	pub const fn new(value:Vec<u8>)->Self{
-		Self(value)
-	}
-}
-impl AsRef<[u8]> for VtxData{
-	fn as_ref(&self)->&[u8]{
-		self.0.as_ref()
-	}
-}
-pub struct VvdData(Vec<u8>);
-impl VvdData{
-	pub const fn new(value:Vec<u8>)->Self{
-		Self(value)
-	}
-}
-impl AsRef<[u8]> for VvdData{
-	fn as_ref(&self)->&[u8]{
-		self.0.as_ref()
-	}
-}
-
-pub struct ModelData{
-	pub mdl:MdlData,
-	pub vtx:VtxData,
-	pub vvd:VvdData,
-}
-impl ModelData{
-	pub fn read_model(&self)->Result<vmdl::Model,vmdl::ModelError>{
-		Ok(vmdl::Model::from_parts(
-			vmdl::mdl::Mdl::read(self.mdl.as_ref())?,
-			vmdl::vtx::Vtx::read(self.vtx.as_ref())?,
-			vmdl::vvd::Vvd::read(self.vvd.as_ref())?,
-		))
-	}
-}

lib/bsp_loader/src/lib.rs

@@ -1,7 +1,17 @@
-mod bsp;
-pub mod data;
+pub mod bsp;
+pub mod mesh;
+mod brush;
 
-pub use data::Bsp;
+const VALVE_SCALE:f32=1.0/16.0;
+pub(crate) fn valve_transform_dist(d:f32)->strafesnet_common::integer::Planar64{
+	(d*VALVE_SCALE).try_into().unwrap()
+}
+pub(crate) fn valve_transform_normal([x,y,z]:[f32;3])->strafesnet_common::integer::Planar64Vec3{
+	strafesnet_common::integer::vec3::try_from_f32_array([x,z,-y]).unwrap()
+}
+pub(crate) fn valve_transform([x,y,z]:[f32;3])->strafesnet_common::integer::Planar64Vec3{
+	strafesnet_common::integer::vec3::try_from_f32_array([x*VALVE_SCALE,z*VALVE_SCALE,-y*VALVE_SCALE]).unwrap()
+}
 
 #[derive(Debug)]
 pub enum ReadError{
@@ -15,6 +25,16 @@ impl std::fmt::Display for ReadError{
 }
 impl std::error::Error for ReadError{}
 
+pub struct Bsp{
+	bsp:vbsp::Bsp,
+	case_folded_file_names:std::collections::HashMap<String,String>,
+}
+impl AsRef<vbsp::Bsp> for Bsp{
+	fn as_ref(&self)->&vbsp::Bsp{
+		&self.bsp
+	}
+}
+
 pub fn read<R:std::io::Read>(mut input:R)->Result<Bsp,ReadError>{
 	let mut s=Vec::new();
 
@@ -23,15 +43,44 @@ pub fn read<R:std::io::Read>(mut input:R)->Result<Bsp,ReadError>{
 
 	vbsp::Bsp::read(s.as_slice()).map(Bsp::new).map_err(ReadError::Bsp)
 }
-
-pub fn convert<AcquireRenderConfigId,AcquireMeshId>(
-	bsp:&Bsp,
-	acquire_render_config_id:AcquireRenderConfigId,
-	acquire_mesh_id:AcquireMeshId
-)->bsp::PartialMap1
-where
-	AcquireRenderConfigId:FnMut(Option<&str>)->strafesnet_common::model::RenderConfigId,
-	AcquireMeshId:FnMut(&str)->strafesnet_common::model::MeshId,
-{
-	bsp::convert_bsp(bsp.as_ref(),acquire_render_config_id,acquire_mesh_id)
-}
+impl Bsp{
+	pub fn new(bsp:vbsp::Bsp)->Self{
+		let case_folded_file_names=bsp.pack.clone().into_zip().lock().unwrap().file_names().map(|s|{
+			(s.to_lowercase(),s.to_owned())
+		}).collect();
+		Self{
+			bsp,
+			case_folded_file_names,
+		}
+	}
+	pub fn pack_get(&self,name_lowercase:&str)->Result<Option<Vec<u8>>,vbsp::BspError>{
+		match self.case_folded_file_names.get(name_lowercase){
+			Some(name_folded)=>self.bsp.pack.get(name_folded),
+			None=>Ok(None),
+		}
+	}
+}
+pub struct Vpk{
+	vpk:vpk::VPK,
+	case_folded_file_names:std::collections::HashMap<String,String>,
+}
+impl AsRef<vpk::VPK> for Vpk{
+	fn as_ref(&self)->&vpk::VPK{
+		&self.vpk
+	}
+}
+impl Vpk{
+	pub fn new(vpk:vpk::VPK)->Vpk{
+		let case_folded_file_names=vpk.tree.keys().map(|s|{
+			(s.to_lowercase(),s.to_owned())
+		}).collect();
+		Vpk{
+			vpk,
+			case_folded_file_names,
+		}
+	}
+	pub fn tree_get(&self,name_lowercase:&str)->Option<&vpk::entry::VPKEntry>{
+		let name_folded=self.case_folded_file_names.get(name_lowercase)?;
+		self.vpk.tree.get(name_folded)
+	}
+}

lib/bsp_loader/src/mesh.rs

@@ -0,0 +1,80 @@
+use std::borrow::Cow;
+
+use strafesnet_common::model;
+use strafesnet_deferred_loader::deferred_loader::RenderConfigDeferredLoader;
+
+use crate::valve_transform;
+
+pub use model::Mesh;
+
+fn ingest_vertex(mb:&mut model::MeshBuilder,vertex:&vmdl::vvd::Vertex,color:model::ColorId)->model::VertexId{
+	let pos=mb.acquire_pos_id(valve_transform(vertex.position.into()));
+	let normal=mb.acquire_normal_id(valve_transform(vertex.normal.into()));
+	let tex=mb.acquire_tex_id(glam::Vec2::from_array(vertex.texture_coordinates));
+	mb.acquire_vertex_id(model::IndexedVertex{
+		pos,
+		tex,
+		normal,
+		color,
+	})
+}
+
+pub fn convert_mesh(model:vmdl::Model,deferred_loader:&mut RenderConfigDeferredLoader<Cow<str>>)->Mesh{
+	let texture_paths=model.texture_directories();
+	if texture_paths.len()!=1{
+		println!("WARNING: multiple texture paths");
+	}
+	let skin=model.skin_tables().nth(0).unwrap();
+
+	let mut mb=model::MeshBuilder::new();
+
+	let color=mb.acquire_color_id(glam::Vec4::ONE);
+
+	let model_vertices=model.vertices();
+
+	let mut graphics_groups=Vec::new();
+	let mut physics_groups=Vec::new();
+	let polygon_groups=model.meshes().enumerate().map(|(polygon_group_id,mesh)|{
+		let polygon_group_id=model::PolygonGroupId::new(polygon_group_id as u32);
+
+		let render_id=if let (Some(texture_path),Some(texture_name))=(texture_paths.get(0),skin.texture(mesh.material_index())){
+			let mut path=std::path::PathBuf::from(texture_path.as_str());
+			path.push(texture_name);
+			let index=path.as_os_str().to_str().map(|s|Cow::Owned(s.to_owned()));
+			deferred_loader.acquire_render_config_id(index)
+		}else{
+			deferred_loader.acquire_render_config_id(None)
+		};
+
+		graphics_groups.push(model::IndexedGraphicsGroup{
+			render:render_id,
+			groups:vec![polygon_group_id],
+		});
+		physics_groups.push(model::IndexedPhysicsGroup{
+			groups:vec![polygon_group_id],
+		});
+		model::PolygonGroup::PolygonList(model::PolygonList::new(
+			//looking at the code, it would seem that the strips are pre-deindexed into triangle lists when calling this function
+			mesh.vertex_strip_indices().flat_map(|mut strip|{
+				std::iter::from_fn(move ||{
+					match (strip.next(),strip.next(),strip.next()){
+						(Some(v1),Some(v2),Some(v3))=>Some([v1,v2,v3]),
+						//ignore extra vertices, not sure what to do in this case, failing the whole conversion could be appropriate
+						_=>None,
+					}
+				})
+			}).filter_map(|[v1,v2,v3]|{
+				// this should probably be a fatal error :D
+				let v1=model_vertices.get(v1)?;
+				let v2=model_vertices.get(v2)?;
+				let v3=model_vertices.get(v3)?;
+				Some(vec![
+					ingest_vertex(&mut mb,v1,color),
+					ingest_vertex(&mut mb,v2,color),
+					ingest_vertex(&mut mb,v3,color),
+				])
+			}).collect()
+		))
+	}).collect();
+	mb.build(polygon_groups,graphics_groups,physics_groups)
+}

lib/common/Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 name = "strafesnet_common"
-version = "0.5.2"
-edition = "2021"
+version = "0.9.0"
+edition = "2024"
 repository = "https://git.itzana.me/StrafesNET/strafe-project"
 license = "MIT OR Apache-2.0"
 description = "Common types and helpers for Strafe Client associated projects."
@@ -12,8 +12,11 @@ authors = ["Rhys Lloyd <krakow20@gmail.com>"]
 [dependencies]
 arrayvec = "0.7.4"
 bitflags = "2.6.0"
-fixed_wide = { path = "../fixed_wide", registry = "strafesnet", features = ["deferred-division","zeroes","wide-mul"] }
-linear_ops = { path = "../linear_ops", registry = "strafesnet", features = ["deferred-division","named-fields"] }
-ratio_ops = { path = "../ratio_ops", registry = "strafesnet" }
-glam = "0.29.0"
+fixed_wide = { workspace = true, features = ["deferred-division","zeroes","wide-mul"] }
+linear_ops = { workspace = true, features = ["deferred-division","named-fields"] }
+ratio_ops = { workspace = true }
+glam.workspace = true
 id = { version = "0.1.0", registry = "strafesnet" }
+
+[lints]
+workspace = true

lib/common/src/aabb.rs

@@ -2,53 +2,70 @@ use crate::integer::{vec3,Planar64Vec3};
 
 #[derive(Clone)]
 pub struct Aabb{
+	// min is inclusive
 	min:Planar64Vec3,
+	// max is not inclusive
 	max:Planar64Vec3,
 }
 
 impl Default for Aabb{
+	#[inline]
 	fn default()->Self{
 		Self{min:vec3::MAX,max:vec3::MIN}
 	}
 }
 
 impl Aabb{
+	#[inline]
 	pub const fn new(min:Planar64Vec3,max:Planar64Vec3)->Self{
 		Self{min,max}
 	}
+	#[inline]
 	pub const fn max(&self)->Planar64Vec3{
 		self.max
 	}
+	#[inline]
 	pub const fn min(&self)->Planar64Vec3{
 		self.min
 	}
+	#[inline]
 	pub fn grow(&mut self,point:Planar64Vec3){
 		self.min=self.min.min(point);
 		self.max=self.max.max(point);
 	}
+	#[inline]
 	pub fn join(&mut self,aabb:&Aabb){
 		self.min=self.min.min(aabb.min);
 		self.max=self.max.max(aabb.max);
 	}
+	#[inline]
 	pub fn inflate(&mut self,hs:Planar64Vec3){
 		self.min-=hs;
 		self.max+=hs;
 	}
+	#[inline]
+	pub fn contains(&self,point:Planar64Vec3)->bool{
+		let bvec=self.min.le(point)&point.lt(self.max);
+		bvec.all()
+	}
+	#[inline]
 	pub fn intersects(&self,aabb:&Aabb)->bool{
 		let bvec=self.min.lt(aabb.max)&aabb.min.lt(self.max);
 		bvec.all()
 	}
+	#[inline]
 	pub fn size(&self)->Planar64Vec3{
 		self.max-self.min
 	}
+	#[inline]
 	pub fn center(&self)->Planar64Vec3{
-		self.min+(self.max-self.min)>>1
+		self.min.map_zip(self.max,|(min,max)|min.midpoint(max))
+	}
+	#[inline]
+	pub fn area_weight(&self)->fixed_wide::types::F128_64{
+		let d=self.max-self.min;
+		d.x*d.y+d.y*d.z+d.z*d.x
 	}
-	//probably use floats for area & volume because we don't care about precision
-	// pub fn area_weight(&self)->f32{
-	// 	let d=self.max-self.min;
-	// 	d.x*d.y+d.y*d.z+d.z*d.x
-	// }
 	// pub fn volume(&self)->f32{
 	// 	let d=self.max-self.min;
 	// 	d.x*d.y*d.z

lib/common/src/bvh.rs

@@ -1,4 +1,10 @@
+use std::cmp::Ordering;
+use std::collections::BTreeMap;
+
 use crate::aabb::Aabb;
+use crate::ray::Ray;
+use crate::integer::{Ratio,Planar64,Planar64Vec3};
+use crate::instruction::{InstructionCollector,TimedInstruction};
 
 //da algaritum
 //lista boxens
@@ -10,35 +16,117 @@ use crate::aabb::Aabb;
 //sort the centerpoints on each axis (3 lists)
 //bv is put into octant based on whether it is upper or lower in each list
 
-pub enum RecursiveContent<R,T>{
-	Branch(Vec<R>),
-	Leaf(T),
+
+pub fn intersect_aabb(ray:&Ray,aabb:&Aabb)->Option<Ratio<Planar64,Planar64>>{
+	// n.(o+d*t)==n.p
+	// n.o + n.d * t == n.p
+	// t == (n.p - n.o)/n.d
+	let mut hit=None;
+	match ray.direction.x.cmp(&Planar64::ZERO){
+		Ordering::Less=>{
+			let rel_min=aabb.min()-ray.origin;
+			let rel_max=aabb.max()-ray.origin;
+			let dy=rel_max.x*ray.direction.y;
+			let dz=rel_max.x*ray.direction.z;
+			// x is negative, so inequalities are flipped
+			if rel_min.y*ray.direction.x>dy&&dy>rel_max.y*ray.direction.x
+			 &&rel_min.z*ray.direction.x>dz&&dz>rel_max.z*ray.direction.x{
+				let t=rel_max.x/ray.direction.x;
+				hit=Some(hit.map_or(t,|best_t|t.min(best_t)));
+			}
+		},
+		Ordering::Equal=>(),
+		Ordering::Greater=>{
+			let rel_min=aabb.min()-ray.origin;
+			let rel_max=aabb.max()-ray.origin;
+			let dy=rel_min.x*ray.direction.y;
+			let dz=rel_min.x*ray.direction.z;
+			// x is positive, so inequalities are normal
+			if rel_min.y*ray.direction.x<dy&&dy<rel_max.y*ray.direction.x
+			 &&rel_min.z*ray.direction.x<dz&&dz<rel_max.z*ray.direction.x{
+				let t=rel_min.x/ray.direction.x;
+				hit=Some(hit.map_or(t,|best_t|t.min(best_t)));
+			}
+		},
+	}
+	match ray.direction.z.cmp(&Planar64::ZERO){
+		Ordering::Less=>{
+			let rel_min=aabb.min()-ray.origin;
+			let rel_max=aabb.max()-ray.origin;
+			let dx=rel_max.z*ray.direction.x;
+			let dy=rel_max.z*ray.direction.y;
+			// z is negative, so inequalities are flipped
+			if rel_min.x*ray.direction.z>dx&&dx>rel_max.x*ray.direction.z
+			 &&rel_min.y*ray.direction.z>dy&&dy>rel_max.y*ray.direction.z{
+				let t=rel_max.z/ray.direction.z;
+				hit=Some(hit.map_or(t,|best_t|t.min(best_t)));
+			}
+		},
+		Ordering::Equal=>(),
+		Ordering::Greater=>{
+			let rel_min=aabb.min()-ray.origin;
+			let rel_max=aabb.max()-ray.origin;
+			let dx=rel_min.z*ray.direction.x;
+			let dy=rel_min.z*ray.direction.y;
+			// z is positive, so inequalities are normal
+			if rel_min.x*ray.direction.z<dx&&dx<rel_max.x*ray.direction.z
+			 &&rel_min.y*ray.direction.z<dy&&dy<rel_max.y*ray.direction.z{
+				let t=rel_min.z/ray.direction.z;
+				hit=Some(hit.map_or(t,|best_t|t.min(best_t)));
+			}
+		},
+	}
+	match ray.direction.y.cmp(&Planar64::ZERO){
+		Ordering::Less=>{
+			let rel_min=aabb.min()-ray.origin;
+			let rel_max=aabb.max()-ray.origin;
+			let dz=rel_max.y*ray.direction.z;
+			let dx=rel_max.y*ray.direction.x;
+			// y is negative, so inequalities are flipped
+			if rel_min.z*ray.direction.y>dz&&dz>rel_max.z*ray.direction.y
+			 &&rel_min.x*ray.direction.y>dx&&dx>rel_max.x*ray.direction.y{
+				let t=rel_max.y/ray.direction.y;
+				hit=Some(hit.map_or(t,|best_t|t.min(best_t)));
+			}
+		},
+		Ordering::Equal=>(),
+		Ordering::Greater=>{
+			let rel_min=aabb.min()-ray.origin;
+			let rel_max=aabb.max()-ray.origin;
+			let dz=rel_min.y*ray.direction.z;
+			let dx=rel_min.y*ray.direction.x;
+			// y is positive, so inequalities are normal
+			if rel_min.z*ray.direction.y<dz&&dz<rel_max.z*ray.direction.y
+			 &&rel_min.x*ray.direction.y<dx&&dx<rel_max.x*ray.direction.y{
+				let t=rel_min.y/ray.direction.y;
+				hit=Some(hit.map_or(t,|best_t|t.min(best_t)));
+			}
+		},
+	}
+	hit
 }
-impl<R,T> Default for RecursiveContent<R,T>{
-	fn default()->Self{
+
+pub enum RecursiveContent<N,L>{
+	Branch(Vec<N>),
+	Leaf(L),
+}
+impl<N,L> RecursiveContent<N,L>{
+	pub fn empty()->Self{
 		Self::Branch(Vec::new())
 	}
 }
-pub struct BvhNode<T>{
-	content:RecursiveContent<BvhNode<T>,T>,
+pub struct BvhNode<L>{
+	content:RecursiveContent<BvhNode<L>,L>,
 	aabb:Aabb,
 }
-impl<T> Default for BvhNode<T>{
-	fn default()->Self{
+impl<L> BvhNode<L>{
+	pub fn empty()->Self{
 		Self{
-			content:Default::default(),
+			content:RecursiveContent::empty(),
 			aabb:Aabb::default(),
 		}
 	}
-}
-pub struct BvhWeightNode<W,T>{
-	content:RecursiveContent<BvhWeightNode<W,T>,T>,
-	weight:W,
-	aabb:Aabb,
-}
-
-impl<T> BvhNode<T>{
-	pub fn the_tester<F:FnMut(&T)>(&self,aabb:&Aabb,f:&mut F){
+	pub fn sample_aabb<F:FnMut(&L)>(&self,aabb:&Aabb,f:&mut F){
 		match &self.content{
 			RecursiveContent::Leaf(model)=>f(model),
 			RecursiveContent::Branch(children)=>for child in children{
@@ -47,51 +135,104 @@ impl<T> BvhNode<T>{
 				//you're probably not going to spend a lot of time outside the map,
 				//so the test is extra work for nothing
 				if aabb.intersects(&child.aabb){
-					child.the_tester(aabb,f);
+					child.sample_aabb(aabb,f);
 				}
 			},
 		}
 	}
-	pub fn into_visitor<F:FnMut(T)>(self,f:&mut F){
-		match self.content{
-			RecursiveContent::Leaf(model)=>f(model),
+	fn populate_nodes<'a,T,IntersectLeaf,IntersectAabb>(
+		&'a self,
+		start_point:Planar64Vec3,
+		collector:&mut InstructionCollector<&'a L,T>,
+		nodes:&mut BTreeMap<T,&'a BvhNode<L>>,
+		start_time:T,
+		intersect_leaf:&IntersectLeaf,
+		intersect_aabb:&IntersectAabb,
+	)
+		where
+			T:Ord+Copy,
+			IntersectLeaf:Fn(&L)->Option<T>,
+			IntersectAabb:Fn(&Aabb)->Option<T>,
+	{
+		match &self.content{
+			RecursiveContent::Leaf(leaf)=>if let Some(time)=intersect_leaf(leaf){
+				let ins=TimedInstruction{time,instruction:leaf};
+				if start_time<ins.time&&ins.time<collector.time(){
+					collector.collect(Some(ins));
+				}
+			},
 			RecursiveContent::Branch(children)=>for child in children{
-				child.into_visitor(f)
-			},
-		}
-	}
-	pub fn weigh_contents<W:Copy+std::iter::Sum<W>,F:Fn(&T)->W>(self,f:&F)->BvhWeightNode<W,T>{
-		match self.content{
-			RecursiveContent::Leaf(model)=>BvhWeightNode{
-				weight:f(&model),
-				content:RecursiveContent::Leaf(model),
-				aabb:self.aabb,
-			},
-			RecursiveContent::Branch(children)=>{
-				let branch:Vec<BvhWeightNode<W,T>>=children.into_iter().map(|child|
-					child.weigh_contents(f)
-				).collect();
-				BvhWeightNode{
-					weight:branch.iter().map(|node|node.weight).sum(),
-					content:RecursiveContent::Branch(branch),
-					aabb:self.aabb,
+				if child.aabb.contains(start_point){
+					child.populate_nodes(start_point,collector,nodes,start_time,intersect_leaf,intersect_aabb);
+				}else{
+					// Am I an upcoming superstar?
+					if let Some(t)=intersect_aabb(&child.aabb){
+						if start_time<t&&t<collector.time(){
+							nodes.insert(t,child);
+						}
+					}
 				}
 			},
 		}
 	}
-}
+	/// Traverse the BVH using the given sampling functions.
+	/// Nodes are tested in order of T returned by IntersectAabb.
+	/// The algorithm ends when T for the next node to test is
+	/// greater than the current best collected T from IntersectLeaf.
+	pub fn traverse<T,IntersectLeaf,IntersectAabb>(
+		&self,
+		start_point:Planar64Vec3,
+		start_time:T,
+		time_limit:T,
+		intersect_leaf:IntersectLeaf,
+		intersect_aabb:IntersectAabb,
+	)->Option<(T,&L)>
+		where
+			T:Ord+Copy,
+			IntersectLeaf:Fn(&L)->Option<T>,
+			IntersectAabb:Fn(&Aabb)->Option<T>,
+	{
+		// source of nondeterminism when Aabb boundaries are coplanar
+		let mut nodes=BTreeMap::new();
 
-impl <W,T> BvhWeightNode<W,T>{
-	pub const fn weight(&self)->&W{
-		&self.weight
+		let mut collector=InstructionCollector::new(time_limit);
+		// break open all nodes that contain ray.origin and populate nodes with future intersection times
+		self.populate_nodes(start_point,&mut collector,&mut nodes,start_time,&intersect_leaf,&intersect_aabb);
+
+		// swim through nodes one at a time
+		while let Some((t,node))=nodes.pop_first(){
+			if collector.time()<t{
+				break;
+			}
+			match &node.content{
+				RecursiveContent::Leaf(leaf)=>if let Some(time)=intersect_leaf(leaf){
+					let ins=TimedInstruction{time:time.into(),instruction:leaf};
+					// this lower bound can also be omitted
+					// but it causes type inference errors lol
+					if start_time<ins.time&&ins.time<collector.time(){
+						collector.collect(Some(ins));
+					}
+				},
+				// break open the node and predict collisions with the child nodes
+				RecursiveContent::Branch(children)=>for child in children{
+					// Am I an upcoming superstar?
+					if let Some(t)=intersect_aabb(&child.aabb){
+						// we don't need to check the lower bound
+						// because child aabbs are guaranteed to be within the parent bounds.
+						if t<collector.time(){
+							nodes.insert(t,child);
+						}
+					}
+				},
+			}
+		}
+
+		collector.take().map(|TimedInstruction{time,instruction:leaf}|(time.into(),leaf))
 	}
-	pub const fn aabb(&self)->&Aabb{
-		&self.aabb
+	pub fn into_inner(self)->(RecursiveContent<BvhNode<L>,L>,Aabb){
+		(self.content,self.aabb)
 	}
-	pub fn into_content(self)->RecursiveContent<BvhWeightNode<W,T>,T>{
-		self.content
-	}
-	pub fn into_visitor<F:FnMut(T)>(self,f:&mut F){
+	pub fn into_visitor<F:FnMut(L)>(self,f:&mut F){
 		match self.content{
 			RecursiveContent::Leaf(model)=>f(model),
 			RecursiveContent::Branch(children)=>for child in children{
@@ -107,18 +248,19 @@ pub fn generate_bvh<T>(boxen:Vec<(T,Aabb)>)->BvhNode<T>{
 
 fn generate_bvh_node<T>(boxen:Vec<(T,Aabb)>,force:bool)->BvhNode<T>{
 	let n=boxen.len();
-	if force||n<20{
-		let mut aabb=Aabb::default();
-		let nodes=boxen.into_iter().map(|b|{
-			aabb.join(&b.1);
+	const MAX_TERMINAL_BRANCH_LEAF_NODES:usize=20;
+	if force||n<MAX_TERMINAL_BRANCH_LEAF_NODES{
+		let mut aabb_outer=Aabb::default();
+		let nodes=boxen.into_iter().map(|(data,aabb)|{
+			aabb_outer.join(&aabb);
 			BvhNode{
-				content:RecursiveContent::Leaf(b.0),
-				aabb:b.1,
+				content:RecursiveContent::Leaf(data),
+				aabb,
 			}
 		}).collect();
 		BvhNode{
 			content:RecursiveContent::Branch(nodes),
-			aabb,
+			aabb:aabb_outer,
 		}
 	}else{
 		let mut sort_x=Vec::with_capacity(n);
@@ -130,13 +272,13 @@ fn generate_bvh_node<T>(boxen:Vec<(T,Aabb)>,force:bool)->BvhNode<T>{
 			sort_y.push((i,center.y));
 			sort_z.push((i,center.z));
 		}
-		sort_x.sort_by(|tup0,tup1|tup0.1.cmp(&tup1.1));
-		sort_y.sort_by(|tup0,tup1|tup0.1.cmp(&tup1.1));
-		sort_z.sort_by(|tup0,tup1|tup0.1.cmp(&tup1.1));
+		sort_x.sort_by_key(|&(_,c)|c);
+		sort_y.sort_by_key(|&(_,c)|c);
+		sort_z.sort_by_key(|&(_,c)|c);
 		let h=n/2;
-		let median_x=sort_x[h].1;
-		let median_y=sort_y[h].1;
-		let median_z=sort_z[h].1;
+		let (_,median_x)=sort_x[h];
+		let (_,median_y)=sort_y[h];
+		let (_,median_z)=sort_z[h];
 		//locate a run of values equal to the median
 		//partition point gives the first index for which the predicate evaluates to false
 		let first_index_eq_median_x=sort_x.partition_point(|&(_,x)|x<median_x);
@@ -175,10 +317,10 @@ fn generate_bvh_node<T>(boxen:Vec<(T,Aabb)>,force:bool)->BvhNode<T>{
 			};
 			list_list[list_id].push((data,aabb));
 		}
-		let mut aabb=Aabb::default();
 		if list_list.len()==1{
 			generate_bvh_node(list_list.remove(0),true)
 		}else{
+			let mut aabb=Aabb::default();
 			BvhNode{
 				content:RecursiveContent::Branch(
 					list_list.into_iter().map(|b|{

lib/common/src/gameplay_attributes.rs

@@ -171,4 +171,7 @@ impl CollisionAttributes{
 	pub fn contact_default()->Self{
 		Self::Contact(ContactAttributes::default())
 	}
+	pub fn intersect_default()->Self{
+		Self::Intersect(IntersectAttributes::default())
+	}
 }

lib/common/src/gameplay_modes.rs

@@ -1,7 +1,6 @@
 use std::collections::{HashSet,HashMap};
 use crate::model::ModelId;
 use crate::gameplay_style;
-use crate::updatable::Updatable;
 
 #[derive(Clone)]
 pub struct StageElement{
@@ -128,18 +127,6 @@ impl Stage{
 		self.unordered_checkpoints.contains(&model_id)
 	}
 }
-#[derive(Default)]
-pub struct StageUpdate{
-	//other behaviour models of this stage can have
-	ordered_checkpoints:HashMap<CheckpointId,ModelId>,
-	unordered_checkpoints:HashSet<ModelId>,
-}
-impl Updatable<StageUpdate> for Stage{
-	fn update(&mut self,update:StageUpdate){
-		self.ordered_checkpoints.extend(update.ordered_checkpoints);
-		self.unordered_checkpoints.extend(update.unordered_checkpoints);
-	}
-}
 
 #[derive(Clone,Copy,Hash,Eq,PartialEq)]
 pub enum Zone{
@@ -153,6 +140,15 @@ impl ModeId{
 	pub const MAIN:Self=Self(0);
 	pub const BONUS:Self=Self(1);
 }
+impl core::fmt::Display for ModeId{
+	fn fmt(&self,f:&mut std::fmt::Formatter<'_>)->core::fmt::Result{
+		match self{
+			&Self::MAIN=>write!(f,"Main"),
+			&Self::BONUS=>write!(f,"Bonus"),
+			&Self(mode_id)=>write!(f,"Bonus{mode_id}"),
+		}
+	}
+}
 #[derive(Clone)]
 pub struct Mode{
 	style:gameplay_style::StyleModifiers,
@@ -211,34 +207,47 @@ impl Mode{
 	pub fn push_stage(&mut self,stage:Stage){
 		self.stages.push(stage)
 	}
-	pub fn get_stage_mut(&mut self,stage:StageId)->Option<&mut Stage>{
-		self.stages.get_mut(stage.0 as usize)
+	pub fn get_stage_mut(&mut self,StageId(stage_id):StageId)->Option<&mut Stage>{
+		self.stages.get_mut(stage_id as usize)
 	}
-	pub fn get_spawn_model_id(&self,stage:StageId)->Option<ModelId>{
-		self.stages.get(stage.0 as usize).map(|s|s.spawn)
+	pub fn get_spawn_model_id(&self,StageId(stage_id):StageId)->Option<ModelId>{
+		self.stages.get(stage_id as usize).map(|s|s.spawn)
 	}
 	pub fn get_zone(&self,model_id:ModelId)->Option<&Zone>{
 		self.zones.get(&model_id)
 	}
-	pub fn get_stage(&self,stage_id:StageId)->Option<&Stage>{
-		self.stages.get(stage_id.0 as usize)
+	pub fn get_stage(&self,StageId(stage_id):StageId)->Option<&Stage>{
+		self.stages.get(stage_id as usize)
 	}
 	pub fn get_element(&self,model_id:ModelId)->Option<&StageElement>{
 		self.elements.get(&model_id)
 	}
+}
+
+#[derive(Clone)]
+pub struct NormalizedMode(Mode);
+impl NormalizedMode{
+	pub fn new(mode:Mode)->Self{
+		Self(mode)
+	}
+	pub fn into_inner(self)->Mode{
+		let Self(mode)=self;
+		mode
+	}
 	//TODO: put this in the SNF
-	pub fn denormalize_data(&mut self){
+	pub fn denormalize(self)->Mode{
+		let NormalizedMode(mut mode)=self;
 		//expand and index normalized data
-		self.zones.insert(self.start,Zone::Start);
-		for (stage_id,stage) in self.stages.iter().enumerate(){
-			self.elements.insert(stage.spawn,StageElement{
+		mode.zones.insert(mode.start,Zone::Start);
+		for (stage_id,stage) in mode.stages.iter().enumerate(){
+			mode.elements.insert(stage.spawn,StageElement{
 				stage_id:StageId(stage_id as u32),
 				force:false,
 				behaviour:StageElementBehaviour::SpawnAt,
 				jump_limit:None,
 			});
 			for (_,&model) in &stage.ordered_checkpoints{
-				self.elements.insert(model,StageElement{
+				mode.elements.insert(model,StageElement{
 					stage_id:StageId(stage_id as u32),
 					force:false,
 					behaviour:StageElementBehaviour::Checkpoint,
@@ -246,7 +255,7 @@ impl Mode{
 				});
 			}
 			for &model in &stage.unordered_checkpoints{
-				self.elements.insert(model,StageElement{
+				mode.elements.insert(model,StageElement{
 					stage_id:StageId(stage_id as u32),
 					force:false,
 					behaviour:StageElementBehaviour::Checkpoint,
@@ -254,53 +263,13 @@ impl Mode{
 				});
 			}
 		}
-	}
-}
-//this would be nice as a macro
-#[derive(Default)]
-pub struct ModeUpdate{
-	zones:HashMap<ModelId,Zone>,
-	stages:HashMap<StageId,StageUpdate>,
-	//mutually exlusive stage element behaviour
-	elements:HashMap<ModelId,StageElement>,
-}
-impl Updatable<ModeUpdate> for Mode{
-	fn update(&mut self,update:ModeUpdate){
-		self.zones.extend(update.zones);
-		for (stage,stage_update) in update.stages{
-			if let Some(stage)=self.stages.get_mut(stage.0 as usize){
-				stage.update(stage_update);
-			}
-		}
-		self.elements.extend(update.elements);
-	}
-}
-impl ModeUpdate{
-	pub fn zone(model_id:ModelId,zone:Zone)->Self{
-		let mut mu=Self::default();
-		mu.zones.insert(model_id,zone);
-		mu
-	}
-	pub fn stage(stage_id:StageId,stage_update:StageUpdate)->Self{
-		let mut mu=Self::default();
-		mu.stages.insert(stage_id,stage_update);
-		mu
-	}
-	pub fn element(model_id:ModelId,element:StageElement)->Self{
-		let mut mu=Self::default();
-		mu.elements.insert(model_id,element);
-		mu
-	}
-	pub fn map_stage_element_ids<F:Fn(StageId)->StageId>(&mut self,f:F){
-		for (_,stage_element) in self.elements.iter_mut(){
-			stage_element.stage_id=f(stage_element.stage_id);
-		}
+		mode
 	}
 }
 
 #[derive(Default,Clone)]
 pub struct Modes{
-	pub modes:Vec<Mode>,
+	modes:Vec<Mode>,
 }
 impl Modes{
 	pub const fn new(modes:Vec<Mode>)->Self{
@@ -314,19 +283,191 @@ impl Modes{
 	pub fn push_mode(&mut self,mode:Mode){
 		self.modes.push(mode)
 	}
-	pub fn get_mode(&self,mode:ModeId)->Option<&Mode>{
-		self.modes.get(mode.0 as usize)
+	pub fn get_mode(&self,ModeId(mode_id):ModeId)->Option<&Mode>{
+		self.modes.get(mode_id as usize)
 	}
 }
-pub struct ModesUpdate{
-	modes:HashMap<ModeId,ModeUpdate>,
+
+#[derive(Clone)]
+pub struct NormalizedModes{
+	modes:Vec<NormalizedMode>,
 }
-impl Updatable<ModesUpdate> for Modes{
-	fn update(&mut self,update:ModesUpdate){
-		for (mode,mode_update) in update.modes{
-			if let Some(mode)=self.modes.get_mut(mode.0 as usize){
-				mode.update(mode_update);
-			}
+impl NormalizedModes{
+	pub fn new(modes:Vec<NormalizedMode>)->Self{
+		Self{modes}
+	}
+	pub fn len(&self)->usize{
+		self.modes.len()
+	}
+	pub fn denormalize(self)->Modes{
+		Modes{
+			modes:self.modes.into_iter().map(NormalizedMode::denormalize).collect(),
 		}
 	}
 }
+impl IntoIterator for NormalizedModes{
+	type Item=<Vec<NormalizedMode> as IntoIterator>::Item;
+	type IntoIter=<Vec<NormalizedMode> as IntoIterator>::IntoIter;
+	fn into_iter(self)->Self::IntoIter{
+		self.modes.into_iter()
+	}
+}
+
+
+#[derive(Default)]
+pub struct StageUpdate{
+	//other behaviour models of this stage can have
+	ordered_checkpoints:HashMap<CheckpointId,ModelId>,
+	unordered_checkpoints:HashSet<ModelId>,
+}
+impl StageUpdate{
+	fn apply_to(self,stage:&mut Stage){
+		stage.ordered_checkpoints.extend(self.ordered_checkpoints);
+		stage.unordered_checkpoints.extend(self.unordered_checkpoints);
+	}
+}
+
+//this would be nice as a macro
+#[derive(Default)]
+pub struct ModeUpdate{
+	zones:Option<(ModelId,Zone)>,
+	stages:Option<(StageId,StageUpdate)>,
+	//mutually exlusive stage element behaviour
+	elements:Option<(ModelId,StageElement)>,
+}
+impl ModeUpdate{
+	fn apply_to(self,mode:&mut Mode){
+		mode.zones.extend(self.zones);
+		if let Some((StageId(stage_id),stage_update))=self.stages{
+			if let Some(stage)=mode.stages.get_mut(stage_id as usize){
+				stage_update.apply_to(stage);
+			}
+		}
+		mode.elements.extend(self.elements);
+	}
+}
+impl ModeUpdate{
+	pub fn zone(model_id:ModelId,zone:Zone)->Self{
+		Self{
+			zones:Some((model_id,zone)),
+			stages:None,
+			elements:None,
+		}
+	}
+	pub fn stage(stage_id:StageId,stage_update:StageUpdate)->Self{
+		Self{
+			zones:None,
+			stages:Some((stage_id,stage_update)),
+			elements:None,
+		}
+	}
+	pub fn element(model_id:ModelId,element:StageElement)->Self{
+		Self{
+			zones:None,
+			stages:None,
+			elements:Some((model_id,element)),
+		}
+	}
+	pub fn map_stage_element_ids<F:Fn(StageId)->StageId>(&mut self,f:F){
+		for (_,stage_element) in self.elements.iter_mut(){
+			stage_element.stage_id=f(stage_element.stage_id);
+		}
+	}
+}
+
+struct ModeBuilder{
+	mode:Mode,
+	stage_id_map:HashMap<StageId,StageId>,
+}
+#[derive(Default)]
+pub struct ModesBuilder{
+	modes:HashMap<ModeId,Mode>,
+	stages:HashMap<ModeId,HashMap<StageId,Stage>>,
+	mode_updates:Vec<(ModeId,ModeUpdate)>,
+	stage_updates:Vec<(ModeId,StageId,StageUpdate)>,
+}
+impl ModesBuilder{
+	pub fn build_normalized(mut self)->NormalizedModes{
+		//collect modes and stages into contiguous arrays
+		let mut unique_modes:Vec<(ModeId,Mode)>
+		=self.modes.into_iter().collect();
+		unique_modes.sort_by_key(|&(mode_id,_)|mode_id);
+		let (mut modes,mode_id_map):(Vec<ModeBuilder>,HashMap<ModeId,ModeId>)
+		=unique_modes.into_iter().enumerate()
+		.map(|(final_mode_id,(builder_mode_id,mut mode))|{
+			(
+				ModeBuilder{
+					stage_id_map:self.stages.remove(&builder_mode_id).map_or_else(||HashMap::new(),|stages|{
+						let mut unique_stages:Vec<(StageId,Stage)>
+						=stages.into_iter().collect();
+						unique_stages.sort_by_key(|&(StageId(stage_id),_)|stage_id);
+						unique_stages.into_iter().enumerate()
+						.map(|(final_stage_id,(builder_stage_id,stage))|{
+							mode.push_stage(stage);
+							(builder_stage_id,StageId::new(final_stage_id as u32))
+						}).collect()
+					}),
+					mode,
+				},
+				(
+					builder_mode_id,
+					ModeId::new(final_mode_id as u32)
+				)
+			)
+		}).unzip();
+		//TODO: failure messages or errors or something
+		//push stage updates
+		for (builder_mode_id,builder_stage_id,stage_update) in self.stage_updates{
+			if let Some(final_mode_id)=mode_id_map.get(&builder_mode_id){
+				if let Some(mode)=modes.get_mut(final_mode_id.get() as usize){
+					if let Some(&final_stage_id)=mode.stage_id_map.get(&builder_stage_id){
+						if let Some(stage)=mode.mode.get_stage_mut(final_stage_id){
+							stage_update.apply_to(stage);
+						}
+					}
+				}
+			}
+		}
+		//push mode updates
+		for (builder_mode_id,mut mode_update) in self.mode_updates{
+			if let Some(final_mode_id)=mode_id_map.get(&builder_mode_id){
+				if let Some(mode)=modes.get_mut(final_mode_id.get() as usize){
+					//map stage id on stage elements
+					mode_update.map_stage_element_ids(|stage_id|
+						//walk down one stage id at a time until a stage is found
+						//TODO use better logic like BTreeMap::upper_bound instead of walking
+						// final_stage_id_from_builder_stage_id.upper_bound(Bound::Included(&stage_id))
+						// .value().copied().unwrap_or(StageId::FIRST)
+						(0..=stage_id.get()).rev().find_map(|builder_stage_id|
+							//map the stage element to that stage
+							mode.stage_id_map.get(&StageId::new(builder_stage_id)).copied()
+						).unwrap_or(StageId::FIRST)
+					);
+					mode_update.apply_to(&mut mode.mode);
+				}
+			}
+		}
+		NormalizedModes::new(modes.into_iter().map(|mode_builder|NormalizedMode(mode_builder.mode)).collect())
+	}
+	pub fn insert_mode(&mut self,mode_id:ModeId,mode:Mode)->Result<(),ExistingEntryError>{
+		error_if_exists(self.modes.insert(mode_id,mode))
+	}
+	pub fn insert_stage(&mut self,mode_id:ModeId,stage_id:StageId,stage:Stage)->Result<(),ExistingEntryError>{
+		error_if_exists(self.stages.entry(mode_id).or_insert(HashMap::new()).insert(stage_id,stage))
+	}
+	pub fn push_mode_update(&mut self,mode_id:ModeId,mode_update:ModeUpdate){
+		self.mode_updates.push((mode_id,mode_update));
+	}
+	// fn push_stage_update(&mut self,mode_id:ModeId,stage_id:StageId,stage_update:StageUpdate){
+	// 	self.stage_updates.push((mode_id,stage_id,stage_update));
+	// }
+}
+
+#[derive(Debug)]
+pub struct ExistingEntryError;
+fn error_if_exists<T>(value:Option<T>)->Result<(),ExistingEntryError>{
+	match value{
+		Some(_)=>Err(ExistingEntryError),
+		None=>Ok(())
+	}
+}

lib/common/src/gameplay_style.rs

@@ -34,7 +34,7 @@ pub struct StyleModifiers{
 	//unused
 	pub mass:Planar64,
 }
-impl std::default::Default for StyleModifiers{
+impl Default for StyleModifiers{
 	fn default()->Self{
 		Self::roblox_bhop()
 	}
@@ -63,22 +63,22 @@ impl JumpImpulse{
 		velocity:Planar64Vec3,
 		jump_dir:Planar64Vec3,
 		gravity:&Planar64Vec3,
-		mass:Planar64,
+		_mass:Planar64,
 	)->Planar64Vec3{
 		match self{
-			&JumpImpulse::Time(time)=>velocity-(*gravity*time).map(|t|t.divide().fix_1()),
+			&JumpImpulse::Time(time)=>velocity-(*gravity*time).map(|t|t.divide().clamp_64()),
 			&JumpImpulse::Height(height)=>{
 				//height==-v.y*v.y/(2*g.y);
 				//use energy to determine max height
 				let gg=gravity.length_squared();
-				let g=gg.sqrt().fix_1();
+				let g=gg.sqrt();
 				let v_g=gravity.dot(velocity);
 				//do it backwards
-				let radicand=v_g*v_g+(g*height*2).fix_4();
-				velocity-(*gravity*(radicand.sqrt().fix_2()+v_g)/gg).divide().fix_1()
+				let radicand=v_g*v_g+(g*height*2).widen_256();
+				velocity-(*gravity*(radicand.sqrt().wrap_128()+v_g)/gg).divide().clamp_64()
 			},
-			&JumpImpulse::Linear(jump_speed)=>velocity+(jump_dir*jump_speed/jump_dir.length()).divide().fix_1(),
-			&JumpImpulse::Energy(energy)=>{
+			&JumpImpulse::Linear(jump_speed)=>velocity+(jump_dir*jump_speed/jump_dir.length()).divide().clamp_64(),
+			&JumpImpulse::Energy(_energy)=>{
 				//calculate energy
 				//let e=gravity.dot(velocity);
 				//add
@@ -91,10 +91,10 @@ impl JumpImpulse{
 	pub fn get_jump_deltav(&self,gravity:&Planar64Vec3,mass:Planar64)->Planar64{
 		//gravity.length() is actually the proper calculation because the jump is always opposite the gravity direction
 		match self{
-			&JumpImpulse::Time(time)=>(gravity.length().fix_1()*time/2).divide().fix_1(),
-			&JumpImpulse::Height(height)=>(gravity.length()*height*2).sqrt().fix_1(),
+			&JumpImpulse::Time(time)=>(gravity.length().wrap_64()*time/2).divide().clamp_64(),
+			&JumpImpulse::Height(height)=>(gravity.length()*height*2).sqrt().wrap_64(),
 			&JumpImpulse::Linear(deltav)=>deltav,
-			&JumpImpulse::Energy(energy)=>(energy.sqrt()*2/mass.sqrt()).divide().fix_1(),
+			&JumpImpulse::Energy(energy)=>(energy.sqrt()*2/mass.sqrt()).divide().clamp_64(),
 		}
 	}
 }
@@ -126,10 +126,10 @@ impl JumpSettings{
 					None=>rel_velocity,
 				};
 				let j=boost_vel.dot(jump_dir);
-				let js=jump_speed.fix_2();
+				let js=jump_speed.widen_128();
 				if j<js{
 					//weak booster: just do a regular jump
-					boost_vel+jump_dir.with_length(js-j).divide().fix_1()
+					boost_vel+jump_dir.with_length(js-j).divide().wrap_64()
 				}else{
 					//activate booster normally, jump does nothing
 					boost_vel
@@ -142,13 +142,13 @@ impl JumpSettings{
 					None=>rel_velocity,
 				};
 				let j=boost_vel.dot(jump_dir);
-				let js=jump_speed.fix_2();
+				let js=jump_speed.widen_128();
 				if j<js{
 					//speed in direction of jump cannot be lower than amount
-					boost_vel+jump_dir.with_length(js-j).divide().fix_1()
+					boost_vel+jump_dir.with_length(js-j).divide().wrap_64()
 				}else{
 					//boost and jump add together
-					boost_vel+jump_dir.with_length(js).divide().fix_1()
+					boost_vel+jump_dir.with_length(js).divide().wrap_64()
 				}
 			}
 			(false,JumpCalculation::Max)=>{
@@ -159,10 +159,10 @@ impl JumpSettings{
 					None=>rel_velocity,
 				};
 				let boost_dot=boost_vel.dot(jump_dir);
-				let js=jump_speed.fix_2();
+				let js=jump_speed.widen_128();
 				if boost_dot<js{
 					//weak boost is extended to jump speed
-					boost_vel+jump_dir.with_length(js-boost_dot).divide().fix_1()
+					boost_vel+jump_dir.with_length(js-boost_dot).divide().wrap_64()
 				}else{
 					//activate booster normally, jump does nothing
 					boost_vel
@@ -174,7 +174,7 @@ impl JumpSettings{
 					Some(booster)=>booster.boost(rel_velocity),
 					None=>rel_velocity,
 				};
-				boost_vel+jump_dir.with_length(jump_speed).divide().fix_1()
+				boost_vel+jump_dir.with_length(jump_speed).divide().wrap_64()
 			},
 		}
 	}
@@ -267,9 +267,9 @@ pub struct StrafeSettings{
 impl StrafeSettings{
 	pub fn tick_velocity(&self,velocity:Planar64Vec3,control_dir:Planar64Vec3)->Option<Planar64Vec3>{
 		let d=velocity.dot(control_dir);
-		let mv=self.mv.fix_2();
+		let mv=self.mv.widen_128();
 		match d<mv{
-			true=>Some(velocity+(control_dir*self.air_accel_limit.map_or(mv-d,|limit|limit.fix_2().min(mv-d))).fix_1()),
+			true=>Some(velocity+(control_dir*self.air_accel_limit.map_or(mv-d,|limit|limit.widen_128().min(mv-d))).wrap_64()),
 			false=>None,
 		}
 	}
@@ -290,7 +290,7 @@ pub struct PropulsionSettings{
 }
 impl PropulsionSettings{
 	pub fn acceleration(&self,control_dir:Planar64Vec3)->Planar64Vec3{
-		(control_dir*self.magnitude).fix_1()
+		(control_dir*self.magnitude).clamp_64()
 	}
 }
 
@@ -310,16 +310,16 @@ pub struct WalkSettings{
 impl WalkSettings{
 	pub fn accel(&self,target_diff:Planar64Vec3,gravity:Planar64Vec3)->Planar64{
 		//TODO: fallible walk accel
-		let diff_len=target_diff.length().fix_1();
+		let diff_len=target_diff.length().wrap_64();
 		let friction=if diff_len<self.accelerate.topspeed{
 			self.static_friction
 		}else{
 			self.kinetic_friction
 		};
-		self.accelerate.accel.min((-gravity.y*friction).fix_1())
+		self.accelerate.accel.min((-gravity.y*friction).clamp_64())
 	}
 	pub fn get_walk_target_velocity(&self,control_dir:Planar64Vec3,normal:Planar64Vec3)->Planar64Vec3{
-		if control_dir==crate::integer::vec3::ZERO{
+		if control_dir==crate::integer::vec3::zero(){
 			return control_dir;
 		}
 		let nn=normal.length_squared();
@@ -329,19 +329,19 @@ impl WalkSettings{
 		let dd=d*d;
 		if dd<nnmm{
 			let cr=normal.cross(control_dir);
-			if cr==crate::integer::vec3::ZERO_2{
-				crate::integer::vec3::ZERO
+			if cr==crate::integer::vec3::zero(){
+				crate::integer::vec3::zero()
 			}else{
-				(cr.cross(normal)*self.accelerate.topspeed/((nn*(nnmm-dd)).sqrt())).divide().fix_1()
+				(cr.cross(normal)*self.accelerate.topspeed/((nn*(nnmm-dd)).sqrt())).divide().clamp_64()
 			}
 		}else{
-			crate::integer::vec3::ZERO
+			crate::integer::vec3::zero()
 		}
 	}
 	pub fn is_slope_walkable(&self,normal:Planar64Vec3,up:Planar64Vec3)->bool{
 		//normal is not guaranteed to be unit length
 		let ny=normal.dot(up);
-		let h=normal.length().fix_1();
+		let h=normal.length().wrap_64();
 		//remember this is a normal vector
 		ny.is_positive()&&h*self.surf_dot<ny
 	}
@@ -355,12 +355,12 @@ pub struct LadderSettings{
 	pub dot:Planar64,
 }
 impl LadderSettings{
-	pub const fn accel(&self,target_diff:Planar64Vec3,gravity:Planar64Vec3)->Planar64{
+	pub const fn accel(&self,_target_diff:Planar64Vec3,_gravity:Planar64Vec3)->Planar64{
 		//TODO: fallible ladder accel
 		self.accelerate.accel
 	}
 	pub fn get_ladder_target_velocity(&self,mut control_dir:Planar64Vec3,normal:Planar64Vec3)->Planar64Vec3{
-		if control_dir==crate::integer::vec3::ZERO{
+		if control_dir==crate::integer::vec3::zero(){
 			return control_dir;
 		}
 		let nn=normal.length_squared();
@@ -368,13 +368,13 @@ impl LadderSettings{
 		let nnmm=nn*mm;
 		let d=normal.dot(control_dir);
 		let mut dd=d*d;
-		if (self.dot*self.dot*nnmm).fix_4()<dd{
+		if (self.dot*self.dot*nnmm).clamp_256()<dd{
 			if d.is_negative(){
-				control_dir=Planar64Vec3::new([Planar64::ZERO,mm.fix_1(),Planar64::ZERO]);
+				control_dir=Planar64Vec3::new([Planar64::ZERO,mm.clamp_64(),Planar64::ZERO]);
 			}else{
-				control_dir=Planar64Vec3::new([Planar64::ZERO,-mm.fix_1(),Planar64::ZERO]);
+				control_dir=Planar64Vec3::new([Planar64::ZERO,-mm.clamp_64(),Planar64::ZERO]);
 			}
-			dd=(normal.y*normal.y).fix_4();
+			dd=(normal.y*normal.y).widen_256();
 		}
 		//n=d if you are standing on top of a ladder and press E.
 		//two fixes:
@@ -382,13 +382,13 @@ impl LadderSettings{
 		//- fix the underlying issue
 		if dd<nnmm{
 			let cr=normal.cross(control_dir);
-			if cr==crate::integer::vec3::ZERO_2{
-				crate::integer::vec3::ZERO
+			if cr==crate::integer::vec3::zero(){
+				crate::integer::vec3::zero()
 			}else{
-				(cr.cross(normal)*self.accelerate.topspeed/((nn*(nnmm-dd)).sqrt())).divide().fix_1()
+				(cr.cross(normal)*self.accelerate.topspeed/((nn*(nnmm-dd)).sqrt())).divide().clamp_64()
 			}
 		}else{
-			crate::integer::vec3::ZERO
+			crate::integer::vec3::zero()
 		}
 	}
 }
@@ -417,7 +417,7 @@ impl Hitbox{
 	}
 	pub fn source()->Self{
 		Self{
-			halfsize:((int3(33,73,33)>>1)*VALVE_SCALE).fix_1(),
+			halfsize:((int3(33,73,33)>>1)*VALVE_SCALE).narrow_64().unwrap(),
 			mesh:HitboxMesh::Box,
 		}
 	}
@@ -529,20 +529,20 @@ impl StyleModifiers{
 
 	pub fn source_bhop()->Self{
 		Self{
-			controls_mask:Controls::all()-Controls::MoveUp-Controls::MoveDown,
+			controls_mask:Controls::all(),
 			controls_mask_state:Controls::all(),
 			strafe:Some(StrafeSettings{
 				enable:ControlsActivation::full_2d(),
-				air_accel_limit:Some(Planar64::raw(150<<28)*100),
-				mv:(Planar64::raw(30)*VALVE_SCALE).fix_1(),
+				air_accel_limit:Some(Planar64::raw((150<<28)*100)),
+				mv:Planar64::raw(30<<28),
 				tick_rate:Ratio64::new(100,AbsoluteTime::ONE_SECOND.get() as u64).unwrap(),
 			}),
 			jump:Some(JumpSettings{
-				impulse:JumpImpulse::Height((int(52)*VALVE_SCALE).fix_1()),
+				impulse:JumpImpulse::Height((int(52)*VALVE_SCALE).narrow_64().unwrap()),
 				calculation:JumpCalculation::JumpThenBoost,
 				limit_minimum:true,
 			}),
-			gravity:(int3(0,-800,0)*VALVE_SCALE).fix_1(),
+			gravity:(int3(0,-800,0)*VALVE_SCALE).narrow_64().unwrap(),
 			mass:int(1),
 			rocket:None,
 			walk:Some(WalkSettings{
@@ -565,25 +565,25 @@ impl StyleModifiers{
 				magnitude:int(12),//?
 			}),
 			hitbox:Hitbox::source(),
-			camera_offset:((int3(0,64,0)-(int3(0,73,0)>>1))*VALVE_SCALE).fix_1(),
+			camera_offset:((int3(0,64,0)-(int3(0,73,0)>>1))*VALVE_SCALE).narrow_64().unwrap(),
 		}
 	}
 	pub fn source_surf()->Self{
 		Self{
-			controls_mask:Controls::all()-Controls::MoveUp-Controls::MoveDown,
+			controls_mask:Controls::all(),
 			controls_mask_state:Controls::all(),
 			strafe:Some(StrafeSettings{
 				enable:ControlsActivation::full_2d(),
-				air_accel_limit:Some((int(150)*66*VALVE_SCALE).fix_1()),
-				mv:(int(30)*VALVE_SCALE).fix_1(),
+				air_accel_limit:Some((int(150)*66*VALVE_SCALE).narrow_64().unwrap()),
+				mv:Planar64::raw(30<<28),
 				tick_rate:Ratio64::new(66,AbsoluteTime::ONE_SECOND.get() as u64).unwrap(),
 			}),
 			jump:Some(JumpSettings{
-				impulse:JumpImpulse::Height((int(52)*VALVE_SCALE).fix_1()),
+				impulse:JumpImpulse::Height((int(52)*VALVE_SCALE).narrow_64().unwrap()),
 				calculation:JumpCalculation::JumpThenBoost,
 				limit_minimum:true,
 			}),
-			gravity:(int3(0,-800,0)*VALVE_SCALE).fix_1(),
+			gravity:(int3(0,-800,0)*VALVE_SCALE).narrow_64().unwrap(),
 			mass:int(1),
 			rocket:None,
 			walk:Some(WalkSettings{
@@ -606,7 +606,7 @@ impl StyleModifiers{
 				magnitude:int(12),//?
 			}),
 			hitbox:Hitbox::source(),
-			camera_offset:((int3(0,64,0)-(int3(0,73,0)>>1))*VALVE_SCALE).fix_1(),
+			camera_offset:((int3(0,64,0)-(int3(0,73,0)>>1))*VALVE_SCALE).narrow_64().unwrap(),
 		}
 	}
 }

lib/common/src/instruction.rs

@@ -1,13 +1,11 @@
-use crate::integer::Time;
-
 #[derive(Clone,Debug)]
 pub struct TimedInstruction<I,T>{
-	pub time:Time<T>,
+	pub time:T,
 	pub instruction:I,
 }
 impl<I,T> TimedInstruction<I,T>{
 	#[inline]
-	pub fn set_time<TimeInner>(self,new_time:Time<TimeInner>)->TimedInstruction<I,TimeInner>{
+	pub fn set_time<T2>(self,new_time:T2)->TimedInstruction<I,T2>{
 		TimedInstruction{
 			time:new_time,
 			instruction:self.instruction,
@@ -17,51 +15,91 @@ impl<I,T> TimedInstruction<I,T>{
 
 /// Ensure all emitted instructions are processed before consuming external instructions
 pub trait InstructionEmitter<I>{
-	type TimeInner;
-	fn next_instruction(&self,time_limit:Time<Self::TimeInner>)->Option<TimedInstruction<I,Self::TimeInner>>;
+	type Time;
+	fn next_instruction(&self,time_limit:Self::Time)->Option<TimedInstruction<I,Self::Time>>;
 }
 /// Apply an atomic state update
 pub trait InstructionConsumer<I>{
-	type TimeInner;
-	fn process_instruction(&mut self,instruction:TimedInstruction<I,Self::TimeInner>);
+	type Time;
+	fn process_instruction(&mut self,instruction:TimedInstruction<I,Self::Time>);
 }
 /// If the object produces its own instructions, allow exhaustively feeding them back in
-pub trait InstructionFeedback<I,T>:InstructionEmitter<I,TimeInner=T>+InstructionConsumer<I,TimeInner=T>
+pub trait InstructionFeedback<I,T>:InstructionEmitter<I,Time=T>+InstructionConsumer<I,Time=T>
 	where
-		Time<T>:Copy,
+		T:Copy,
 {
 	#[inline]
-	fn process_exhaustive(&mut self,time_limit:Time<T>){
+	fn process_exhaustive(&mut self,time_limit:T){
 		while let Some(instruction)=self.next_instruction(time_limit){
 			self.process_instruction(instruction);
 		}
 	}
+	#[inline]
+	fn into_iter(self,time_limit:T)->InstructionIter<I,T,Self>
+		where
+			Self:Sized
+	{
+		InstructionIter{
+			time_limit,
+			feedback:self,
+			_phantom:core::marker::PhantomData,
+		}
+	}
 }
-impl<I,T,X> InstructionFeedback<I,T> for X
+impl<I,T,F> InstructionFeedback<I,T> for F
 	where
-		Time<T>:Copy,
-		X:InstructionEmitter<I,TimeInner=T>+InstructionConsumer<I,TimeInner=T>,
+		T:Copy,
+		F:InstructionEmitter<I,Time=T>+InstructionConsumer<I,Time=T>,
 {}
+pub struct InstructionIter<I,T:Copy,F:InstructionFeedback<I,T>>{
+	time_limit:T,
+	feedback:F,
+	_phantom:core::marker::PhantomData<I>,
+}
+impl<I,T,F> Iterator for InstructionIter<I,T,F>
+	where
+		I:Clone,
+		T:Clone+Copy,
+		F:InstructionFeedback<I,T>,
+{
+	type Item=TimedInstruction<I,T>;
+	fn next(&mut self)->Option<Self::Item>{
+		let instruction=self.feedback.next_instruction(self.time_limit)?;
+		self.feedback.process_instruction(instruction.clone());
+		Some(instruction)
+	}
+}
 
 //PROPER PRIVATE FIELDS!!!
 pub struct InstructionCollector<I,T>{
-	time:Time<T>,
+	time:T,
 	instruction:Option<I>,
 }
-impl<I,T> InstructionCollector<I,T>
-	where Time<T>:Copy+PartialOrd,
-{
+impl<I,T> InstructionCollector<I,T>{
 	#[inline]
-	pub const fn new(time:Time<T>)->Self{
+	pub const fn new(time:T)->Self{
 		Self{
 			time,
 			instruction:None
 		}
 	}
 	#[inline]
-	pub const fn time(&self)->Time<T>{
+	pub fn take(self)->Option<TimedInstruction<I,T>>{
+		//STEAL INSTRUCTION AND DESTROY INSTRUCTIONCOLLECTOR
+		self.instruction.map(|instruction|TimedInstruction{
+			time:self.time,
+			instruction
+		})
+	}
+}
+impl<I,T:Copy> InstructionCollector<I,T>{
+	#[inline]
+	pub const fn time(&self)->T{
 		self.time
 	}
+}
+impl<I,T:PartialOrd> InstructionCollector<I,T>{
+	#[inline]
 	pub fn collect(&mut self,instruction:Option<TimedInstruction<I,T>>){
 		if let Some(ins)=instruction{
 			if ins.time<self.time{
@@ -70,11 +108,4 @@ impl<I,T> InstructionCollector<I,T>
 			}
 		}
 	}
-	pub fn instruction(self)->Option<TimedInstruction<I,T>>{
-		//STEAL INSTRUCTION AND DESTROY INSTRUCTIONCOLLECTOR
-		self.instruction.map(|instruction|TimedInstruction{
-			time:self.time,
-			instruction
-		})
-	}
 }

lib/common/src/integer.rs

@@ -1,19 +1,26 @@
-pub use fixed_wide::fixed::{Fixed,Fix};
-pub use ratio_ops::ratio::{Ratio,Divide};
+pub use fixed_wide::fixed::*;
+pub use ratio_ops::ratio::{Ratio,Divide,Parity};
+
+pub mod fixed_types{
+	pub use fixed_wide::types::*;
+}
+
+use fixed_wide::types::F128_64;
 
 //integer units
 
 /// specific example of a "default" time type
-#[derive(Clone,Copy,Hash,Eq,PartialEq,PartialOrd,Debug)]
+#[derive(Clone,Copy,Hash,Eq,PartialEq,Ord,PartialOrd,Debug)]
 pub enum TimeInner{}
 pub type AbsoluteTime=Time<TimeInner>;
 
-#[derive(Clone,Copy,Hash,Eq,PartialEq,PartialOrd,Debug)]
+#[derive(Clone,Copy,Hash,Eq,PartialEq,Ord,PartialOrd,Debug)]
 pub struct Time<T>(i64,core::marker::PhantomData<T>);
 impl<T> Time<T>{
 	pub const MIN:Self=Self::raw(i64::MIN);
 	pub const MAX:Self=Self::raw(i64::MAX);
 	pub const ZERO:Self=Self::raw(0);
+	pub const EPSILON:Self=Self::raw(1);
 	pub const ONE_SECOND:Self=Self::raw(1_000_000_000);
 	pub const ONE_MILLISECOND:Self=Self::raw(1_000_000);
 	pub const ONE_MICROSECOND:Self=Self::raw(1_000);
@@ -55,22 +62,36 @@ impl<T> Time<T>{
 	pub const fn coerce<U>(self)->Time<U>{
 		Time::raw(self.0)
 	}
+	#[inline]
+	pub const fn div_euclid(self,other:Self)->Self{
+		Time::raw(self.0.div_euclid(other.0))
+	}
+	#[inline]
+	pub const fn rem_euclid(self,other:Self)->Self{
+		Time::raw(self.0.rem_euclid(other.0))
+	}
 }
 impl<T> From<Planar64> for Time<T>{
 	#[inline]
 	fn from(value:Planar64)->Self{
-		Self::raw((value*Planar64::raw(1_000_000_000)).fix_1().to_raw())
+		Self::raw((value*Planar64::raw(1_000_000_000)).clamp_64().to_raw())
+	}
+}
+impl<T> From<Time<T>> for Ratio<Planar64,Planar64>{
+	#[inline]
+	fn from(value:Time<T>)->Self{
+		value.to_ratio()
 	}
 }
 impl<T,Num,Den,N1,T1> From<Ratio<Num,Den>> for Time<T>
 	where
 		Num:core::ops::Mul<Planar64,Output=N1>,
 		N1:Divide<Den,Output=T1>,
-		T1:Fix<Planar64>,
+		T1:Clamp<Planar64>,
 {
 	#[inline]
 	fn from(value:Ratio<Num,Den>)->Self{
-		Self::raw((value*Planar64::raw(1_000_000_000)).divide().fix().to_raw())
+		Self::raw((value*Planar64::raw(1_000_000_000)).divide().clamp().to_raw())
 	}
 }
 impl<T> std::fmt::Display for Time<T>{
@@ -79,7 +100,7 @@ impl<T> std::fmt::Display for Time<T>{
 		write!(f,"{}s+{:09}ns",self.0/Self::ONE_SECOND.0,self.0%Self::ONE_SECOND.0)
 	}
 }
-impl<T> std::default::Default for Time<T>{
+impl<T> Default for Time<T>{
 	fn default()->Self{
 		Self::raw(0)
 	}
@@ -119,36 +140,51 @@ impl_time_additive_assign_operator!(core::ops::AddAssign,add_assign);
 impl_time_additive_assign_operator!(core::ops::SubAssign,sub_assign);
 impl_time_additive_assign_operator!(core::ops::RemAssign,rem_assign);
 impl<T> std::ops::Mul for Time<T>{
-	type Output=Ratio<fixed_wide::fixed::Fixed<2,64>,fixed_wide::fixed::Fixed<2,64>>;
+	type Output=Ratio<F128_64,F128_64>;
 	#[inline]
 	fn mul(self,rhs:Self)->Self::Output{
-		Ratio::new(Fixed::raw(self.0)*Fixed::raw(rhs.0),Fixed::raw_digit(1_000_000_000i64.pow(2)))
+		Ratio::new(Fixed::raw(self.0)*Fixed::raw(rhs.0),Fixed::from_u64(1_000_000_000u64.pow(2)))
 	}
 }
-impl<T> std::ops::Div<i64> for Time<T>{
-	type Output=Self;
-	#[inline]
-	fn div(self,rhs:i64)->Self::Output{
-		Self::raw(self.0/rhs)
-	}
-}
-impl<T> std::ops::Mul<i64> for Time<T>{
-	type Output=Self;
-	#[inline]
-	fn mul(self,rhs:i64)->Self::Output{
-		Self::raw(self.0*rhs)
+macro_rules! impl_time_i64_rhs_operator {
+	($op:ident,$method:ident)=>{
+		impl<T> core::ops::$op<i64> for Time<T>{
+			type Output=Self;
+			#[inline]
+			fn $method(self,rhs:i64)->Self::Output{
+				Self::raw(self.0.$method(rhs))
+			}
+		}
 	}
 }
+impl_time_i64_rhs_operator!(Div,div);
+impl_time_i64_rhs_operator!(Mul,mul);
+impl_time_i64_rhs_operator!(Shr,shr);
+impl_time_i64_rhs_operator!(Shl,shl);
 impl<T> core::ops::Mul<Time<T>> for Planar64{
-	type Output=Ratio<Fixed<2,64>,Planar64>;
+	type Output=Ratio<F128_64,Planar64>;
+	#[inline]
 	fn mul(self,rhs:Time<T>)->Self::Output{
 		Ratio::new(self*Fixed::raw(rhs.0),Planar64::raw(1_000_000_000))
 	}
 }
+impl<T> From<Time<T>> for f32{
+	#[inline]
+	fn from(value:Time<T>)->Self{
+		value.get() as f32/Time::<T>::ONE_SECOND.get() as f32
+	}
+}
+impl<T> From<Time<T>> for f64{
+	#[inline]
+	fn from(value:Time<T>)->Self{
+		value.get() as f64/Time::<T>::ONE_SECOND.get() as f64
+	}
+}
 #[cfg(test)]
 mod test_time{
 	use super::*;
-	type Time=super::AbsoluteTime;
+	use fixed_wide::types::F64_32;
+	type Time=AbsoluteTime;
 	#[test]
 	fn time_from_planar64(){
 		let a:Time=Planar64::from(1).into();
@@ -162,13 +198,13 @@ mod test_time{
 	#[test]
 	fn time_squared(){
 		let a=Time::from_secs(2);
-		assert_eq!(a*a,Ratio::new(Fixed::<2,64>::raw_digit(1_000_000_000i64.pow(2))*4,Fixed::<2,64>::raw_digit(1_000_000_000i64.pow(2))));
+		assert_eq!(a*a,Ratio::new(F128_64::from_u64(1_000_000_000u64.pow(2))*4,F128_64::from_u64(1_000_000_000u64.pow(2))));
 	}
 	#[test]
 	fn time_times_planar64(){
 		let a=Time::from_secs(2);
 		let b=Planar64::from(2);
-		assert_eq!(b*a,Ratio::new(Fixed::<2,64>::raw_digit(1_000_000_000*(1<<32))<<2,Fixed::<1,32>::raw_digit(1_000_000_000)));
+		assert_eq!(b*a,Ratio::new(F128_64::from_u64(1_000_000_000*(1<<32))<<2,F64_32::from_u64(1_000_000_000)));
 	}
 }
 
@@ -206,11 +242,11 @@ impl Ratio64{
 	}
 	#[inline]
 	pub const fn mul_int(&self,rhs:i64)->i64{
-		rhs*self.num/(self.den as i64)
+		(rhs as i128*self.num as i128/self.den as i128) as i64
 	}
 	#[inline]
 	pub const fn rhs_div_int(&self,rhs:i64)->i64{
-		rhs*(self.den as i64)/self.num
+		(rhs as i128*self.den as i128/self.num as i128) as i64
 	}
 	#[inline]
 	pub const fn mul_ref(&self,rhs:&Ratio64)->Ratio64{
@@ -255,7 +291,6 @@ fn integer_decode_f64(f: f64) -> (u64, i16, i8) {
 pub enum Ratio64TryFromFloatError{
 	Nan,
 	Infinite,
-	Subnormal,
 	HighlyNegativeExponent(i16),
 	HighlyPositiveExponent(i16),
 }
@@ -290,8 +325,10 @@ fn ratio64_from_mes((m,e,s):(u64,i16,i8))->Result<Ratio64,Ratio64TryFromFloatErr
 
 		Ok(Ratio64::new(num as i64,den as u64).unwrap())
 	}else if e<0{
+		// simple exact representation
 		Ok(Ratio64::new((m as i64)*(s as i64),1<<-e).unwrap())
 	}else if (64-m.leading_zeros() as i16)+e<64{
+		// integer
 		Ok(Ratio64::new((m as i64)*(s as i64)*(1<<e),1).unwrap())
 	}else{
 		Err(Ratio64TryFromFloatError::HighlyPositiveExponent(e))
@@ -323,6 +360,29 @@ impl TryFrom<f64> for Ratio64{
 		}
 	}
 }
+#[cfg(test)]
+fn req(r0:Ratio64,r1:Ratio64){
+	println!("r0={r0:?} r1={r1:?}");
+	assert_eq!(r0.num(),r1.num(),"Nums not eq");
+	assert_eq!(r0.den(),r1.den(),"Dens not eq");
+}
+#[test]
+fn test_ratio64_from_float(){
+	req(2.0.try_into().unwrap(),Ratio64::new(2,1).unwrap());
+	req(1.0.try_into().unwrap(),Ratio64::new(1,1).unwrap());
+	req(0.5.try_into().unwrap(),Ratio64::new(1,2).unwrap());
+	req(1.1.try_into().unwrap(),Ratio64::new(2476979795053773,2251799813685248).unwrap());
+	req(0.8.try_into().unwrap(),Ratio64::new(3602879701896397,4503599627370496).unwrap());
+	req(0.61.try_into().unwrap(),Ratio64::new(5494391545392005,9007199254740992).unwrap());
+	req(0.01.try_into().unwrap(),Ratio64::new(5764607523034235,576460752303423488).unwrap());
+	req(0.001.try_into().unwrap(),Ratio64::new(1152921504606847,1152921504606846976).unwrap());
+	req(0.00001.try_into().unwrap(),Ratio64::new(89605456633725,8960545663372499267).unwrap());
+	req(0.00000000001.try_into().unwrap(),Ratio64::new(35204848,3520484800000000213).unwrap());
+	req(0.000000000000000001.try_into().unwrap(),Ratio64::new(11,10999999999999999213).unwrap());
+	req(2222222222222.0.try_into().unwrap(),Ratio64::new(2222222222222,1).unwrap());
+	req(core::f64::consts::PI.try_into().unwrap(),Ratio64::new(884279719003555,281474976710656).unwrap());
+}
+
 impl std::ops::Mul<Ratio64> for Ratio64{
 	type Output=Ratio64;
 	#[inline]
@@ -394,6 +454,10 @@ impl Angle32{
 	pub const NEG_FRAC_PI_2:Self=Self(-1<<30);
 	pub const PI:Self=Self(-1<<31);
 	#[inline]
+	pub const fn raw(num:i32)->Self{
+		Self(num)
+	}
+	#[inline]
 	pub const fn wrap_from_i64(theta:i64)->Self{
 		//take lower bits
 		//note: this was checked on compiler explorer and compiles to 1 instruction!
@@ -508,8 +572,8 @@ fn angle_sin_cos(){
 		println!("cordic s={} c={}",(s/h).divide(),(c/h).divide());
 		let (fs,fc)=f.sin_cos();
 		println!("float s={} c={}",fs,fc);
-		assert!(close_enough((c/h).divide().fix_1(),Planar64::raw((fc*((1u64<<32) as f64)) as i64)));
-		assert!(close_enough((s/h).divide().fix_1(),Planar64::raw((fs*((1u64<<32) as f64)) as i64)));
+		assert!(close_enough((c/h).divide().wrap_64(),Planar64::raw((fc*((1u64<<32) as f64)) as i64)));
+		assert!(close_enough((s/h).divide().wrap_64(),Planar64::raw((fs*((1u64<<32) as f64)) as i64)));
 	}
 	test_angle(1.0);
 	test_angle(std::f64::consts::PI/4.0);
@@ -540,8 +604,8 @@ impl TryFrom<[f32;3]> for Unit32Vec3{
 }
 */
 
-pub type Planar64TryFromFloatError=fixed_wide::fixed::FixedFromFloatError;
-pub type Planar64=fixed_wide::types::I32F32;
+pub type Planar64TryFromFloatError=FixedFromFloatError;
+pub type Planar64=fixed_wide::types::F64_32;
 pub type Planar64Vec3=linear_ops::types::Vector3<Planar64>;
 pub type Planar64Mat3=linear_ops::types::Matrix3<Planar64>;
 pub mod vec3{
@@ -549,8 +613,6 @@ pub mod vec3{
 	pub use linear_ops::types::Vector3;
 	pub const MIN:Planar64Vec3=Planar64Vec3::new([Planar64::MIN;3]);
 	pub const MAX:Planar64Vec3=Planar64Vec3::new([Planar64::MAX;3]);
-	pub const ZERO:Planar64Vec3=Planar64Vec3::new([Planar64::ZERO;3]);
-	pub const ZERO_2:linear_ops::types::Vector3<Fixed::<2,64>>=linear_ops::types::Vector3::new([Fixed::<2,64>::ZERO;3]);
 	pub const X:Planar64Vec3=Planar64Vec3::new([Planar64::ONE,Planar64::ZERO,Planar64::ZERO]);
 	pub const Y:Planar64Vec3=Planar64Vec3::new([Planar64::ZERO,Planar64::ONE,Planar64::ZERO]);
 	pub const Z:Planar64Vec3=Planar64Vec3::new([Planar64::ZERO,Planar64::ZERO,Planar64::ONE]);
@@ -559,6 +621,10 @@ pub mod vec3{
 	pub const NEG_Y:Planar64Vec3=Planar64Vec3::new([Planar64::ZERO,Planar64::NEG_ONE,Planar64::ZERO]);
 	pub const NEG_Z:Planar64Vec3=Planar64Vec3::new([Planar64::ZERO,Planar64::ZERO,Planar64::NEG_ONE]);
 	pub const NEG_ONE:Planar64Vec3=Planar64Vec3::new([Planar64::NEG_ONE,Planar64::NEG_ONE,Planar64::NEG_ONE]);
+	// TODO: use #![feature(generic_const_items)] when stabilized https://github.com/rust-lang/rust/issues/113521
+	pub const fn zero<const N:usize,const F:usize>()->Vector3<Fixed<N,F>>{
+		Vector3::new([Fixed::ZERO;3])
+	}
 	#[inline]
 	pub const fn int(x:i32,y:i32,z:i32)->Planar64Vec3{
 		Planar64Vec3::new([Planar64::raw((x as i64)<<32),Planar64::raw((y as i64)<<32),Planar64::raw((z as i64)<<32)])
@@ -618,8 +684,8 @@ pub mod mat3{
 		let (yc,ys)=y.cos_sin();
 		Planar64Mat3::from_cols([
 			Planar64Vec3::new([xc,Planar64::ZERO,-xs]),
-			Planar64Vec3::new([(xs*ys).fix_1(),yc,(xc*ys).fix_1()]),
-			Planar64Vec3::new([(xs*yc).fix_1(),-ys,(xc*yc).fix_1()]),
+			Planar64Vec3::new([(xs*ys).wrap_64(),yc,(xc*ys).wrap_64()]),
+			Planar64Vec3::new([(xs*yc).wrap_64(),-ys,(xc*yc).wrap_64()]),
 		])
 	}
 	#[inline]
@@ -641,19 +707,27 @@ pub mod mat3{
 	}
 }
 
-#[derive(Clone,Copy,Default,Hash,Eq,PartialEq)]
+#[derive(Clone,Copy,Debug,Default,Hash,Eq,PartialEq)]
 pub struct Planar64Affine3{
 	pub matrix3:Planar64Mat3,//includes scale above 1
 	pub translation:Planar64Vec3,
 }
 impl Planar64Affine3{
+	pub const IDENTITY:Self=Self::new(mat3::identity(),vec3::zero());
 	#[inline]
 	pub const fn new(matrix3:Planar64Mat3,translation:Planar64Vec3)->Self{
 		Self{matrix3,translation}
 	}
 	#[inline]
-	pub fn transform_point3(&self,point:Planar64Vec3)->vec3::Vector3<Fixed<2,64>>{
-		self.translation.fix_2()+self.matrix3*point
+	pub const fn from_translation(translation:Planar64Vec3)->Self{
+		Self{
+			matrix3:mat3::identity(),
+			translation,
+		}
+	}
+	#[inline]
+	pub fn transform_point3(&self,point:Planar64Vec3)->vec3::Vector3<F128_64>{
+		self.translation.widen_128()+self.matrix3*point
 	}
 }
 impl Into<glam::Mat4> for Planar64Affine3{

lib/common/src/lib.rs

@@ -1,5 +1,6 @@
 pub mod bvh;
 pub mod map;
+pub mod ray;
 pub mod run;
 pub mod aabb;
 pub mod model;
@@ -8,7 +9,6 @@ pub mod timer;
 pub mod integer;
 pub mod physics;
 pub mod session;
-pub mod updatable;
 pub mod instruction;
 pub mod gameplay_attributes;
 pub mod gameplay_modes;

lib/common/src/map.rs

@@ -4,7 +4,7 @@ use crate::gameplay_attributes;
 //this is a temporary struct to try to get the code running again
 //TODO: use snf::map::Region to update the data in physics and graphics instead of this
 pub struct CompleteMap{
-	pub modes:gameplay_modes::Modes,
+	pub modes:gameplay_modes::NormalizedModes,
 	pub attributes:Vec<gameplay_attributes::CollisionAttributes>,
 	pub meshes:Vec<model::Mesh>,
 	pub models:Vec<model::Model>,

lib/common/src/model.rs

@@ -1,4 +1,6 @@
-use crate::integer::{Planar64Vec3,Planar64Affine3};
+use std::collections::HashMap;
+
+use crate::integer::{Planar64,Planar64Vec3,Planar64Affine3};
 use crate::gameplay_attributes;
 
 pub type TextureCoordinate=glam::Vec2;
@@ -123,6 +125,92 @@ pub struct Mesh{
 	pub physics_groups:Vec<IndexedPhysicsGroup>,
 }
 
+#[derive(Default)]
+pub struct MeshBuilder{
+	unique_pos:Vec<Planar64Vec3>,//Unit32Vec3
+	unique_normal:Vec<Planar64Vec3>,//Unit32Vec3
+	unique_tex:Vec<TextureCoordinate>,
+	unique_color:Vec<Color4>,
+	unique_vertices:Vec<IndexedVertex>,
+	pos_id_from:HashMap<Planar64Vec3,PositionId>,//Unit32Vec3
+	normal_id_from:HashMap<Planar64Vec3,NormalId>,//Unit32Vec3
+	tex_id_from:HashMap<[u32;2],TextureCoordinateId>,
+	color_id_from:HashMap<[u32;4],ColorId>,
+	vertex_id_from:HashMap<IndexedVertex,VertexId>,
+}
+impl MeshBuilder{
+	pub fn new()->Self{
+		Self::default()
+	}
+	pub fn build(
+		self,
+		polygon_groups:Vec<PolygonGroup>,
+		graphics_groups:Vec<IndexedGraphicsGroup>,
+		physics_groups:Vec<IndexedPhysicsGroup>,
+	)->Mesh{
+		let MeshBuilder{
+			unique_pos,
+			unique_normal,
+			unique_tex,
+			unique_color,
+			unique_vertices,
+			..
+		}=self;
+		Mesh{
+			unique_pos,
+			unique_normal,
+			unique_tex,
+			unique_color,
+			unique_vertices,
+			polygon_groups,
+			graphics_groups,
+			physics_groups,
+		}
+	}
+	pub fn acquire_pos_id(&mut self,pos:Planar64Vec3)->PositionId{
+		// Truncate the 16 most precise bits of the vertex positions.
+		// This allows the normal vectors to exactly represent the face.
+		// Remove this in Mesh V2
+		const MASK:Planar64=Planar64::raw(!((1<<16)-1));
+		let pos=pos.map(|c|c&MASK);
+		*self.pos_id_from.entry(pos).or_insert_with(||{
+			let pos_id=PositionId::new(self.unique_pos.len() as u32);
+			self.unique_pos.push(pos);
+			pos_id
+		})
+	}
+	pub fn acquire_normal_id(&mut self,normal:Planar64Vec3)->NormalId{
+		*self.normal_id_from.entry(normal).or_insert_with(||{
+			let normal_id=NormalId::new(self.unique_normal.len() as u32);
+			self.unique_normal.push(normal);
+			normal_id
+		})
+	}
+	pub fn acquire_tex_id(&mut self,tex:TextureCoordinate)->TextureCoordinateId{
+		let h=tex.to_array().map(f32::to_bits);
+		*self.tex_id_from.entry(h).or_insert_with(||{
+			let tex_id=TextureCoordinateId::new(self.unique_tex.len() as u32);
+			self.unique_tex.push(tex);
+			tex_id
+		})
+	}
+	pub fn acquire_color_id(&mut self,color:Color4)->ColorId{
+		let h=color.to_array().map(f32::to_bits);
+		*self.color_id_from.entry(h).or_insert_with(||{
+			let color_id=ColorId::new(self.unique_color.len() as u32);
+			self.unique_color.push(color);
+			color_id
+		})
+	}
+	pub fn acquire_vertex_id(&mut self,vertex:IndexedVertex)->VertexId{
+		*self.vertex_id_from.entry(vertex.clone()).or_insert_with(||{
+			let vertex_id=VertexId::new(self.unique_vertices.len() as u32);
+			self.unique_vertices.push(vertex);
+			vertex_id
+		})
+	}
+}
+
 #[derive(Debug,Clone,Copy,Hash,id::Id,Eq,PartialEq)]
 pub struct ModelId(u32);
 pub struct Model{

lib/common/src/physics.rs

@@ -1,7 +1,7 @@
 use crate::mouse::MouseState;
 use crate::gameplay_modes::{ModeId,StageId};
 
-#[derive(Clone,Copy,Hash,Eq,PartialEq,PartialOrd,Debug)]
+#[derive(Clone,Copy,Hash,Eq,PartialEq,Ord,PartialOrd,Debug)]
 pub enum TimeInner{}
 pub type Time=crate::integer::Time<TimeInner>;
 
@@ -36,6 +36,7 @@ pub enum SetControlInstruction{
 	SetMoveForward(bool),
 	SetJump(bool),
 	SetZoom(bool),
+	SetSprint(bool),
 }
 #[derive(Clone,Debug)]
 pub enum ModeInstruction{

lib/common/src/ray.rs

@@ -0,0 +1,20 @@
+use ratio_ops::ratio::Ratio;
+use crate::integer::{self,Planar64,Planar64Vec3};
+
+pub struct Ray{
+	pub origin:Planar64Vec3,
+	pub direction:Planar64Vec3,
+}
+impl Ray{
+	pub fn extrapolate<Num,Den,N1,T1>(&self,t:Ratio<Num,Den>)->Planar64Vec3
+	where
+		Num:Copy,
+		Den:Copy,
+		Num:core::ops::Mul<Planar64,Output=N1>,
+		Planar64:core::ops::Mul<Den,Output=N1>,
+		N1:integer::Divide<Den,Output=T1>,
+		T1:integer::Clamp<Planar64>,
+	{
+		self.origin+self.direction.map(|elem|(t*elem).divide().clamp())
+	}
+}

lib/common/src/run.rs

@@ -2,7 +2,7 @@ use crate::timer::{TimerFixed,Realtime,Paused,Unpaused};
 
 use crate::physics::{TimeInner as PhysicsTimeInner,Time as PhysicsTime};
 
-#[derive(Clone,Copy,Hash,Eq,PartialEq,PartialOrd,Debug)]
+#[derive(Clone,Copy,Hash,Eq,PartialEq,Ord,PartialOrd,Debug)]
 pub enum TimeInner{}
 pub type Time=crate::integer::Time<TimeInner>;
 
@@ -53,36 +53,23 @@ impl std::fmt::Display for Error{
 impl std::error::Error for Error{}
 
 #[derive(Clone,Copy,Debug)]
-enum RunState{
+pub enum RunState{
 	Created,
 	Started{timer:TimerFixed<Realtime<PhysicsTimeInner,TimeInner>,Unpaused>},
 	Finished{timer:TimerFixed<Realtime<PhysicsTimeInner,TimeInner>,Paused>},
 }
-
-#[derive(Clone,Copy,Debug)]
-pub struct Run{
-	state:RunState,
-	flagged:Option<FlagReason>,
-}
-
-impl Run{
-	pub fn new()->Self{
-		Self{
-			state:RunState::Created,
-			flagged:None,
-		}
-	}
+impl RunState{
 	pub fn time(&self,time:PhysicsTime)->Time{
-		match &self.state{
+		match &self{
 			RunState::Created=>Time::ZERO,
 			RunState::Started{timer}=>timer.time(time),
-			RunState::Finished{timer}=>timer.time(time),
+			RunState::Finished{timer}=>timer.time(),
 		}
 	}
 	pub fn start(&mut self,time:PhysicsTime)->Result<(),Error>{
-		match &self.state{
+		match &self{
 			RunState::Created=>{
-				self.state=RunState::Started{
+				*self=RunState::Started{
 					timer:TimerFixed::new(time,Time::ZERO),
 				};
 				Ok(())
@@ -93,10 +80,10 @@ impl Run{
 	}
 	pub fn finish(&mut self,time:PhysicsTime)->Result<(),Error>{
 		//this uses Copy
-		match &self.state{
+		match &self{
 			RunState::Created=>Err(Error::NotStarted),
 			RunState::Started{timer}=>{
-				self.state=RunState::Finished{
+				*self=RunState::Finished{
 					timer:timer.into_paused(time),
 				};
 				Ok(())
@@ -104,10 +91,43 @@ impl Run{
 			RunState::Finished{..}=>Err(Error::AlreadyFinished),
 		}
 	}
+}
+
+#[derive(Clone,Copy,Debug)]
+pub struct Run{
+	state:RunState,
+	flag_reason:Option<FlagReason>,
+}
+
+impl Run{
+	pub fn new()->Self{
+		Self{
+			state:RunState::Created,
+			flag_reason:None,
+		}
+	}
+	pub fn time(&self,time:PhysicsTime)->Time{
+		self.state.time(time)
+	}
+	pub fn start(&mut self,time:PhysicsTime)->Result<(),Error>{
+		self.state.start(time)
+	}
+	pub fn finish(&mut self,time:PhysicsTime)->Result<(),Error>{
+		self.state.finish(time)
+	}
 	pub fn flag(&mut self,flag_reason:FlagReason){
 		//don't replace the first reason the run was flagged
-		if self.flagged.is_none(){
-			self.flagged=Some(flag_reason);
+		if self.flag_reason.is_none(){
+			self.flag_reason=Some(flag_reason);
+		}
+	}
+	pub fn flag_reason(&self)->Option<FlagReason>{
+		self.flag_reason
+	}
+	pub fn get_finish_time(&self)->Option<Time>{
+		match &self.state{
+			RunState::Finished{timer}=>Some(timer.time()),
+			_=>None,
 		}
 	}
 }

lib/common/src/session.rs

@@ -1,3 +1,3 @@
-#[derive(Clone,Copy,Hash,Eq,PartialEq,PartialOrd,Debug)]
+#[derive(Clone,Copy,Hash,Eq,PartialEq,Ord,PartialOrd,Debug)]
 pub enum TimeInner{}
 pub type Time=crate::integer::Time<TimeInner>;

lib/common/src/timer.rs

@@ -23,7 +23,7 @@ impl PauseState for Unpaused{
 }
 
 #[derive(Clone,Copy,Hash,Eq,PartialEq,PartialOrd,Debug)]
-enum Inner{}
+pub enum Inner{}
 type InnerTime=Time<Inner>;
 
 #[derive(Clone,Copy,Debug)]
@@ -69,10 +69,8 @@ impl<In,Out> Scaled<In,Out>
 	const fn get_scale(&self)->Ratio64{
 		self.scale
 	}
-	fn set_scale(&mut self,time:Time<In>,new_scale:Ratio64){
-		let new_time=self.get_time(time);
+	const fn set_scale(&mut self,new_scale:Ratio64){
 		self.scale=new_scale;
-		self.set_time(time,new_time);
 	}
 }
 
@@ -132,7 +130,7 @@ pub struct TimerFixed<T:TimerState,P:PauseState>{
 	_paused:P,
 }
 
-//scaled timer methods are generic across PauseState
+//some scaled timer methods are generic across PauseState
 impl<P:PauseState,In,Out> TimerFixed<Scaled<In,Out>,P>
 	where Time<In>:Copy,
 {
@@ -147,8 +145,22 @@ impl<P:PauseState,In,Out> TimerFixed<Scaled<In,Out>,P>
 	pub const fn get_scale(&self)->Ratio64{
 		self.state.get_scale()
 	}
+}
+// setting the scale of an unpaused timer is different than a paused timer
+impl<In,Out> TimerFixed<Scaled<In,Out>,Unpaused>
+	where Time<In>:Copy,
+{
 	pub fn set_scale(&mut self,time:Time<In>,new_scale:Ratio64){
-		self.state.set_scale(time,new_scale)
+		let new_time=self.state.get_time(time);
+		self.state.set_scale(new_scale);
+		self.state.set_time(time,new_time);
+	}
+}
+impl<In,Out> TimerFixed<Scaled<In,Out>,Paused>
+	where Time<In>:Copy,
+{
+	pub fn set_scale(&mut self,new_scale:Ratio64){
+		self.state.set_scale(new_scale);
 	}
 }
 
@@ -157,7 +169,7 @@ impl<T:TimerState> TimerFixed<T,Paused>
 	where Time<T::In>:Copy,
 {
 	pub fn into_unpaused(self,time:Time<T::In>)->TimerFixed<T,Unpaused>{
-		let new_time=self.time(time);
+		let new_time=self.time();
 		let mut timer=TimerFixed{
 			state:self.state,
 			_paused:Unpaused,
@@ -165,6 +177,9 @@ impl<T:TimerState> TimerFixed<T,Paused>
 		timer.set_time(time,new_time);
 		timer
 	}
+	pub fn time(&self)->Time<T::Out>{
+		self.state.get_offset().coerce()
+	}
 }
 impl<T:TimerState> TimerFixed<T,Unpaused>
 	where Time<T::In>:Copy,
@@ -178,6 +193,9 @@ impl<T:TimerState> TimerFixed<T,Unpaused>
 		timer.set_time(time,new_time);
 		timer
 	}
+	pub fn time(&self,time:Time<T::In>)->Time<T::Out>{
+		self.state.get_time(time)
+	}
 }
 
 //the new constructor and time queries are generic across both
@@ -199,12 +217,6 @@ impl<T:TimerState,P:PauseState> TimerFixed<T,P>{
 	pub fn into_state(self)->T{
 		self.state
 	}
-	pub fn time(&self,time:Time<T::In>)->Time<T::Out>{
-		match P::IS_PAUSED{
-			true=>self.state.get_offset().coerce(),
-			false=>self.state.get_time(time),
-		}
-	}
 	pub fn set_time(&mut self,time:Time<T::In>,new_time:Time<T::Out>){
 		match P::IS_PAUSED{
 			true=>self.state.set_offset(new_time.coerce()),
@@ -256,7 +268,7 @@ impl<T:TimerState> Timer<T>
 	}
 	pub fn time(&self,time:Time<T::In>)->Time<T::Out>{
 		match self{
-			Self::Paused(timer)=>timer.time(time),
+			Self::Paused(timer)=>timer.time(),
 			Self::Unpaused(timer)=>timer.time(time),
 		}
 	}
@@ -305,7 +317,7 @@ impl<In,Out> Timer<Scaled<In,Out>>
 	}
 	pub fn set_scale(&mut self,time:Time<In>,new_scale:Ratio64){
 		match self{
-			Self::Paused(timer)=>timer.set_scale(time,new_scale),
+			Self::Paused(timer)=>timer.set_scale(new_scale),
 			Self::Unpaused(timer)=>timer.set_scale(time,new_scale),
 		}
 	}
@@ -329,7 +341,7 @@ mod test{
 		//create a paused timer that reads 0s
 		let timer=TimerFixed::<Scaled<Parent,Calculated>,Paused>::from_state(Scaled::new(0.5f32.try_into().unwrap(),sec!(0)));
 		//the paused timer at 1 second should read 0s
-		assert_eq!(timer.time(sec!(1)),sec!(0));
+		assert_eq!(timer.time(),sec!(0));
 
 		//unpause it after one second
 		let timer=timer.into_unpaused(sec!(1));
@@ -339,7 +351,7 @@ mod test{
 		//pause the timer after 11 seconds
 		let timer=timer.into_paused(sec!(11));
 		//the paused timer at 20 seconds should read 5s
-		assert_eq!(timer.time(sec!(20)),sec!(5));
+		assert_eq!(timer.time(),sec!(5));
 	}
 	#[test]
 	fn test_timer()->Result<(),Error>{

lib/common/src/updatable.rs

@@ -1,56 +0,0 @@
-pub trait Updatable<Updater>{
-	fn update(&mut self,update:Updater);
-}
-#[derive(Clone,Copy,Hash,Eq,PartialEq)]
-struct InnerId(u32);
-#[derive(Clone)]
-struct Inner{
-	id:InnerId,
-	enabled:bool,
-}
-#[derive(Clone,Copy,Hash,Eq,PartialEq)]
-struct OuterId(u32);
-struct Outer{
-	id:OuterId,
-	inners:std::collections::HashMap<InnerId,Inner>,
-}
-
-enum Update<I,U>{
-	Insert(I),
-	Update(U),
-	Remove
-}
-
-struct InnerUpdate{
-	//#[updatable(Update)]
-	enabled:Option<bool>,
-}
-struct OuterUpdate{
-	//#[updatable(Insert,Update,Remove)]
-	inners:std::collections::HashMap<InnerId,Update<Inner,InnerUpdate>>,
-	//#[updatable(Update)]
-	//inners:std::collections::HashMap<InnerId,InnerUpdate>,
-}
-impl Updatable<InnerUpdate> for Inner{
-	fn update(&mut self,update:InnerUpdate){
-		if let Some(enabled)=update.enabled{
-			self.enabled=enabled;
-		}
-	}
-}
-impl Updatable<OuterUpdate> for Outer{
-	fn update(&mut self,update:OuterUpdate){
-		for (id,up) in update.inners{
-			match up{
-				Update::Insert(new_inner)=>self.inners.insert(id,new_inner),
-				Update::Update(inner_update)=>self.inners.get_mut(&id).map(|inner|{
-					let old=inner.clone();
-					inner.update(inner_update);
-					old
-				}),
-				Update::Remove=>self.inners.remove(&id),
-			};
-		}
-	}
-}
-//*/

lib/deferred_loader/Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 name = "strafesnet_deferred_loader"
-version = "0.4.1"
-edition = "2021"
+version = "0.6.0"
+edition = "2024"
 repository = "https://git.itzana.me/StrafesNET/strafe-project"
 license = "MIT OR Apache-2.0"
 description = "Acquire IDs for objects before loading them in bulk."
@@ -9,13 +9,8 @@ authors = ["Rhys Lloyd <krakow20@gmail.com>"]
 
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 
-[features]
-default = ["legacy"]
-legacy = ["dep:url","dep:vbsp"]
-#roblox = ["dep:lazy-regex"]
-#source = ["dep:vbsp"]
-
 [dependencies]
-strafesnet_common = { path = "../common", registry = "strafesnet" }
-url = { version = "2.5.2", optional = true }
-vbsp = { version = "0.6.0", optional = true }
+strafesnet_common.workspace = true
+
+[lints]
+workspace = true

lib/deferred_loader/src/deferred_loader.rs

@@ -0,0 +1,116 @@
+use std::collections::HashMap;
+use crate::loader::Loader;
+use crate::mesh::Meshes;
+use crate::texture::{RenderConfigs,Texture};
+use strafesnet_common::model::{MeshId,RenderConfig,RenderConfigId,TextureId};
+
+#[derive(Clone,Copy,Debug)]
+pub enum LoadFailureMode{
+	DefaultToNone,
+	Fatal,
+}
+
+pub struct RenderConfigDeferredLoader<H>{
+	texture_count:u32,
+	render_configs:Vec<RenderConfig>,
+	render_config_id_from_asset_id:HashMap<Option<H>,RenderConfigId>,
+}
+impl<H> RenderConfigDeferredLoader<H>{
+	pub fn new()->Self{
+		Self{
+			texture_count:0,
+			render_configs:Vec::new(),
+			render_config_id_from_asset_id:HashMap::new(),
+		}
+	}
+}
+
+impl<H:core::hash::Hash+Eq> RenderConfigDeferredLoader<H>{
+	pub fn acquire_render_config_id(&mut self,index:Option<H>)->RenderConfigId{
+		let some_texture=index.is_some();
+		*self.render_config_id_from_asset_id.entry(index).or_insert_with(||{
+			//create the render config.
+			let render_config=if some_texture{
+				let render_config=RenderConfig::texture(TextureId::new(self.texture_count));
+				self.texture_count+=1;
+				render_config
+			}else{
+				RenderConfig::default()
+			};
+			let render_id=RenderConfigId::new(self.render_configs.len() as u32);
+			self.render_configs.push(render_config);
+			render_id
+		})
+	}
+	pub fn into_indices(self)->impl Iterator<Item=H>{
+		self.render_config_id_from_asset_id.into_keys().flatten()
+	}
+	pub fn into_render_configs<'a,L:Loader<Resource=Texture,Index<'a>=H>+'a>(mut self,loader:&mut L,failure_mode:LoadFailureMode)->Result<RenderConfigs,L::Error>{
+		let mut sorted_textures=vec![None;self.texture_count as usize];
+		for (index_option,render_config_id) in self.render_config_id_from_asset_id{
+			let render_config=&mut self.render_configs[render_config_id.get() as usize];
+			if let (Some(index),Some(texture_id))=(index_option,render_config.texture){
+				let resource_result=loader.load(index);
+				let texture=match failure_mode{
+					// if texture fails to load, use no texture
+					LoadFailureMode::DefaultToNone=>match resource_result{
+						Ok(texture)=>Some(texture),
+						Err(e)=>{
+							render_config.texture=None;
+							println!("Error loading texture: {e}");
+							None
+						},
+					},
+					// loading failure is fatal
+					LoadFailureMode::Fatal=>Some(resource_result?)
+				};
+				sorted_textures[texture_id.get() as usize]=texture;
+			}
+		}
+		Ok(RenderConfigs::new(
+			sorted_textures,
+			self.render_configs,
+		))
+	}
+}
+
+pub struct MeshDeferredLoader<H>{
+	mesh_id_from_asset_id:HashMap<H,MeshId>,
+}
+impl<H> MeshDeferredLoader<H>{
+	pub fn new()->Self{
+		Self{
+			mesh_id_from_asset_id:HashMap::new(),
+		}
+	}
+}
+
+impl<H:core::hash::Hash+Eq> MeshDeferredLoader<H>{
+	pub fn acquire_mesh_id(&mut self,index:H)->MeshId{
+		let mesh_id=MeshId::new(self.mesh_id_from_asset_id.len() as u32);
+		*self.mesh_id_from_asset_id.entry(index).or_insert(mesh_id)
+	}
+	pub fn into_indices(self)->impl Iterator<Item=H>{
+		self.mesh_id_from_asset_id.into_keys()
+	}
+	pub fn into_meshes<'a,M:Clone,L:Loader<Resource=M,Index<'a>=H>+'a>(self,loader:&mut L,failure_mode:LoadFailureMode)->Result<Meshes<M>,L::Error>{
+		let mut mesh_list=vec![None;self.mesh_id_from_asset_id.len()];
+		for (index,mesh_id) in self.mesh_id_from_asset_id{
+			let resource_result=loader.load(index);
+			let mesh=match failure_mode{
+				// if mesh fails to load, use no mesh
+				LoadFailureMode::DefaultToNone=>match resource_result{
+					Ok(mesh)=>Some(mesh),
+					Err(e)=>{
+						println!("Error loading mesh: {e}");
+						None
+					},
+				},
+				// loading failure is fatal
+				LoadFailureMode::Fatal=>Some(resource_result?)
+			};
+			mesh_list[mesh_id.get() as usize]=mesh;
+		}
+		Ok(Meshes::new(mesh_list))
+	}
+}

lib/deferred_loader/src/lib.rs

@@ -1,34 +1,5 @@
-#[cfg(feature="legacy")]
-mod roblox_legacy;
-#[cfg(feature="legacy")]
-mod source_legacy;
-#[cfg(feature="roblox")]
-mod roblox;
-#[cfg(feature="source")]
-mod source;
-
-#[cfg(any(feature="roblox",feature="legacy"))]
-pub mod rbxassetid;
 
+pub mod mesh;
+pub mod loader;
 pub mod texture;
-#[cfg(any(feature="source",feature="legacy"))]
-pub mod valve_mesh;
-#[cfg(any(feature="roblox",feature="legacy"))]
-pub mod roblox_mesh;
-
-#[cfg(feature="legacy")]
-pub fn roblox_legacy()->roblox_legacy::Loader{
-	roblox_legacy::Loader::new()
-}
-#[cfg(feature="legacy")]
-pub fn source_legacy()->source_legacy::Loader{
-	source_legacy::Loader::new()
-}
-#[cfg(feature="roblox")]
-pub fn roblox()->roblox::Loader{
-	roblox::Loader::new()
-}
-#[cfg(feature="source")]
-pub fn source()->source::Loader{
-	source::Loader::new()
-}
+pub mod deferred_loader;

lib/deferred_loader/src/loader.rs

@@ -0,0 +1,8 @@
+use std::error::Error;
+
+pub trait Loader{
+	type Error:Error;
+	type Index<'a> where Self:'a;
+	type Resource;
+	fn load(&mut self,index:Self::Index<'_>)->Result<Self::Resource,Self::Error>;
+}

lib/deferred_loader/src/mesh.rs

@@ -0,0 +1,17 @@
+use strafesnet_common::model::MeshId;
+
+pub struct Meshes<M>{
+	meshes:Vec<Option<M>>,
+}
+impl<M> Meshes<M>{
+	pub(crate) const fn new(meshes:Vec<Option<M>>)->Self{
+		Self{
+			meshes,
+		}
+	}
+	pub fn consume(self)->impl Iterator<Item=(MeshId,M)>{
+		self.meshes.into_iter().enumerate().filter_map(|(mesh_id,maybe_mesh)|
+			maybe_mesh.map(|mesh|(MeshId::new(mesh_id as u32),mesh))
+		)
+	}
+}

lib/deferred_loader/src/rbxassetid.rs

@@ -1,48 +0,0 @@
-#[derive(Hash,Eq,PartialEq)]
-pub struct RobloxAssetId(pub u64);
-#[derive(Debug)]
-#[allow(dead_code)]
-pub struct StringWithError{
-	string:String,
-	error:RobloxAssetIdParseErr,
-}
-impl std::fmt::Display for StringWithError{
-	fn fmt(&self,f:&mut std::fmt::Formatter<'_>)->std::fmt::Result{
-		write!(f,"{self:?}")
-	}
-}
-impl std::error::Error for StringWithError{}
-impl StringWithError{
-	const fn new(
-		string:String,
-		error:RobloxAssetIdParseErr,
-	)->Self{
-		Self{string,error}
-	}
-}
-#[derive(Debug)]
-pub enum RobloxAssetIdParseErr{
-	Url(url::ParseError),
-	UnknownScheme,
-	ParseInt(std::num::ParseIntError),
-	MissingAssetId,
-}
-impl std::str::FromStr for RobloxAssetId{
-	type Err=StringWithError;
-	fn from_str(s:&str)->Result<Self,Self::Err>{
-		let url=url::Url::parse(s).map_err(|e|StringWithError::new(s.to_owned(),RobloxAssetIdParseErr::Url(e)))?;
-		let parsed_asset_id=match url.scheme(){
-			"rbxassetid"=>url.domain().ok_or_else(||StringWithError::new(s.to_owned(),RobloxAssetIdParseErr::MissingAssetId))?.parse(),
-			"http"|"https"=>{
-				let (_,asset_id)=url.query_pairs()
-				.find(|(id,_)|match id.as_ref(){
-					"ID"|"id"|"Id"|"iD"=>true,
-					_=>false,
-				}).ok_or_else(||StringWithError::new(s.to_owned(),RobloxAssetIdParseErr::MissingAssetId))?;
-				asset_id.parse()
-			},
-			_=>Err(StringWithError::new(s.to_owned(),RobloxAssetIdParseErr::UnknownScheme))?,
-		};
-		Ok(Self(parsed_asset_id.map_err(|e|StringWithError::new(s.to_owned(),RobloxAssetIdParseErr::ParseInt(e)))?))
-	}
-}

lib/deferred_loader/src/roblox_legacy.rs

@@ -1,112 +0,0 @@
-use std::io::Read;
-use std::collections::HashMap;
-use crate::roblox_mesh;
-use crate::texture::{RenderConfigs,Texture};
-use strafesnet_common::model::{MeshId,RenderConfig,RenderConfigId,TextureId};
-use crate::rbxassetid::RobloxAssetId;
-
-#[derive(Default)]
-pub struct RenderConfigLoader{
-	texture_count:u32,
-	render_configs:Vec<RenderConfig>,
-	render_config_id_from_asset_id:HashMap<Option<RobloxAssetId>,RenderConfigId>,
-}
-
-impl RenderConfigLoader{
-	pub fn acquire_render_config_id(&mut self,name:Option<&str>)->RenderConfigId{
-		let render_id=RenderConfigId::new(self.render_config_id_from_asset_id.len() as u32);
-		let index=name.and_then(|name|{
-			match name.parse::<RobloxAssetId>(){
-				Ok(asset_id)=>Some(asset_id),
-				Err(e)=>{
-					println!("Failed to parse AssetId: {e}");
-					None
-				},
-			}
-		});
-		*self.render_config_id_from_asset_id.entry(index).or_insert_with(||{
-			//create the render config.
-			let render_config=if name.is_some(){
-				let render_config=RenderConfig::texture(TextureId::new(self.texture_count));
-				self.texture_count+=1;
-				render_config
-			}else{
-				RenderConfig::default()
-			};
-			self.render_configs.push(render_config);
-			render_id
-		})
-	}
-}
-
-#[derive(Default)]
-pub struct MeshLoader{
-	mesh_id_from_asset_id:HashMap<Option<RobloxAssetId>,MeshId>,
-}
-
-impl MeshLoader{
-	pub fn acquire_mesh_id(&mut self,name:&str)->MeshId{
-		let mesh_id=MeshId::new(self.mesh_id_from_asset_id.len() as u32);
-		let index=match name.parse::<RobloxAssetId>(){
-			Ok(asset_id)=>Some(asset_id),
-			Err(e)=>{
-				println!("Failed to parse AssetId: {e}");
-				None
-			},
-		};
-		*self.mesh_id_from_asset_id.entry(index).or_insert(mesh_id)
-	}
-	pub fn load_meshes(&mut self)->Result<roblox_mesh::Meshes,std::io::Error>{
-		let mut mesh_data=vec![None;self.mesh_id_from_asset_id.len()];
-		for (asset_id_option,mesh_id) in &self.mesh_id_from_asset_id{
-			if let Some(asset_id)=asset_id_option{
-				if let Ok(mut file)=std::fs::File::open(format!("meshes/{}",asset_id.0)){
-					//TODO: parallel
-					let mut data=Vec::<u8>::new();
-					file.read_to_end(&mut data)?;
-					mesh_data[mesh_id.get() as usize]=Some(roblox_mesh::RobloxMeshData::new(data));
-				}else{
-					println!("[roblox_legacy] no mesh name={}",asset_id.0);
-				}
-			}
-		}
-		Ok(roblox_mesh::Meshes::new(mesh_data))
-	}
-}
-
-pub struct Loader{
-	render_config_loader:RenderConfigLoader,
-	mesh_loader:MeshLoader,
-}
-impl Loader{
-	pub fn new()->Self{
-		Self{
-			render_config_loader:RenderConfigLoader::default(),
-			mesh_loader:MeshLoader::default(),
-		}
-	}
-	pub fn get_inner_mut(&mut self)->(&mut RenderConfigLoader,&mut MeshLoader){
-		(&mut self.render_config_loader,&mut self.mesh_loader)
-	}
-	pub fn into_render_configs(mut self)->Result<RenderConfigs,std::io::Error>{
-		let mut sorted_textures=vec![None;self.render_config_loader.texture_count as usize];
-		for (asset_id_option,render_config_id) in self.render_config_loader.render_config_id_from_asset_id{
-			let render_config=self.render_config_loader.render_configs.get_mut(render_config_id.get() as usize).unwrap();
-			if let (Some(asset_id),Some(texture_id))=(asset_id_option,render_config.texture){
-				if let Ok(mut file)=std::fs::File::open(format!("textures/{}.dds",asset_id.0)){
-					//TODO: parallel
-					let mut data=Vec::<u8>::new();
-					file.read_to_end(&mut data)?;
-					sorted_textures[texture_id.get() as usize]=Some(Texture::ImageDDS(data));
-				}else{
-					//texture failed to load
-					render_config.texture=None;
-				}
-			}
-		}
-		Ok(RenderConfigs::new(
-			sorted_textures,
-			self.render_config_loader.render_configs,
-		))
-	}
-}

lib/deferred_loader/src/roblox_mesh.rs

@@ -1,30 +0,0 @@
-use strafesnet_common::model::MeshId;
-
-#[derive(Clone)]
-pub struct RobloxMeshData(Vec<u8>);
-impl RobloxMeshData{
-	pub(crate) fn new(data:Vec<u8>)->Self{
-		Self(data)
-	}
-	pub fn get(self)->Vec<u8>{
-		self.0
-	}
-}
-pub struct Meshes{
-	meshes:Vec<Option<RobloxMeshData>>,
-}
-impl Meshes{
-	pub(crate) const fn new(meshes:Vec<Option<RobloxMeshData>>)->Self{
-		Self{
-			meshes,
-		}
-	}
-	pub fn get_texture(&self,texture_id:MeshId)->Option<&RobloxMeshData>{
-		self.meshes.get(texture_id.get() as usize)?.as_ref()
-	}
-	pub fn into_iter(self)->impl Iterator<Item=(MeshId,RobloxMeshData)>{
-		self.meshes.into_iter().enumerate().filter_map(|(mesh_id,maybe_mesh)|
-			maybe_mesh.map(|mesh|(MeshId::new(mesh_id as u32),mesh))
-		)
-	}
-}

lib/deferred_loader/src/source_legacy.rs

@@ -1,102 +0,0 @@
-use std::io::Read;
-use std::collections::HashMap;
-use crate::valve_mesh;
-use crate::texture::{Texture,RenderConfigs};
-use strafesnet_common::model::{MeshId,TextureId,RenderConfig,RenderConfigId};
-
-pub struct RenderConfigLoader{
-	texture_count:u32,
-	render_configs:Vec<RenderConfig>,
-	texture_paths:HashMap<Option<Box<str>>,RenderConfigId>,
-}
-impl RenderConfigLoader{
-	pub fn acquire_render_config_id(&mut self,name:Option<&str>)->RenderConfigId{
-		let render_id=RenderConfigId::new(self.texture_paths.len() as u32);
-		*self.texture_paths.entry(name.map(Into::into)).or_insert_with(||{
-			//create the render config.
-			let render_config=if name.is_some(){
-				let render_config=RenderConfig::texture(TextureId::new(self.texture_count));
-				self.texture_count+=1;
-				render_config
-			}else{
-				RenderConfig::default()
-			};
-			self.render_configs.push(render_config);
-			render_id
-		})
-	}
-}
-pub struct MeshLoader{
-	mesh_paths:HashMap<Box<str>,MeshId>,
-}
-impl MeshLoader{
-	pub fn acquire_mesh_id(&mut self,name:&str)->MeshId{
-		let mesh_id=MeshId::new(self.mesh_paths.len() as u32);
-		*self.mesh_paths.entry(name.into()).or_insert(mesh_id)
-	}
-	//load_meshes should look like load_textures
-	pub fn load_meshes(&mut self,bsp:&vbsp::Bsp)->valve_mesh::Meshes{
-		let mut mesh_data=vec![None;self.mesh_paths.len()];
-		for (mesh_path,mesh_id) in &self.mesh_paths{
-			let mesh_path_lower=mesh_path.to_lowercase();
-			//.mdl, .vvd, .dx90.vtx
-			let path=std::path::PathBuf::from(mesh_path_lower.as_str());
-			let mut vvd_path=path.clone();
-			let mut vtx_path=path.clone();
-			vvd_path.set_extension("vvd");
-			vtx_path.set_extension("dx90.vtx");
-			match (bsp.pack.get(mesh_path_lower.as_str()),bsp.pack.get(vvd_path.as_os_str().to_str().unwrap()),bsp.pack.get(vtx_path.as_os_str().to_str().unwrap())){
-				(Ok(Some(mdl_file)),Ok(Some(vvd_file)),Ok(Some(vtx_file)))=>{
-					mesh_data[mesh_id.get() as usize]=Some(valve_mesh::ModelData{
-						mdl:valve_mesh::MdlData::new(mdl_file),
-						vtx:valve_mesh::VtxData::new(vtx_file),
-						vvd:valve_mesh::VvdData::new(vvd_file),
-					});
-				},
-				_=>println!("no model name={}",mesh_path),
-			}
-		}
-		valve_mesh::Meshes::new(mesh_data)
-	}
-}
-
-pub struct Loader{
-	render_config_loader:RenderConfigLoader,
-	mesh_loader:MeshLoader,
-}
-impl Loader{
-	pub fn new()->Self{
-		Self{
-			render_config_loader:RenderConfigLoader{
-				texture_count:0,
-				texture_paths:HashMap::new(),
-				render_configs:Vec::new(),
-			},
-			mesh_loader:MeshLoader{mesh_paths:HashMap::new()},
-		}
-	}
-	pub fn get_inner_mut(&mut self)->(&mut RenderConfigLoader,&mut MeshLoader){
-		(&mut self.render_config_loader,&mut self.mesh_loader)
-	}
-	pub fn into_render_configs(mut self)->Result<RenderConfigs,std::io::Error>{
-		let mut sorted_textures=vec![None;self.render_config_loader.texture_count as usize];
-		for (texture_path,render_config_id) in self.render_config_loader.texture_paths{
-			let render_config=self.render_config_loader.render_configs.get_mut(render_config_id.get() as usize).unwrap();
-			if let (Some(texture_path),Some(texture_id))=(texture_path,render_config.texture){
-				if let Ok(mut file)=std::fs::File::open(format!("textures/{}.dds",texture_path)){
-					//TODO: parallel
-					let mut data=Vec::<u8>::new();
-					file.read_to_end(&mut data)?;
-					sorted_textures[texture_id.get() as usize]=Some(Texture::ImageDDS(data));
-				}else{
-					//texture failed to load
-					render_config.texture=None;
-				}
-			}
-		}
-		Ok(RenderConfigs::new(
-			sorted_textures,
-			self.render_config_loader.render_configs,
-		))
-	}
-}

lib/deferred_loader/src/valve_mesh.rs

@@ -1,60 +0,0 @@
-use strafesnet_common::model::MeshId;
-
-//duplicate this code for now
-#[derive(Clone)]
-pub struct MdlData(Vec<u8>);
-impl MdlData{
-	pub const fn new(value:Vec<u8>)->Self{
-		Self(value)
-	}
-	pub fn get(self)->Vec<u8>{
-		self.0
-	}
-}
-#[derive(Clone)]
-pub struct VtxData(Vec<u8>);
-impl VtxData{
-	pub const fn new(value:Vec<u8>)->Self{
-		Self(value)
-	}
-	pub fn get(self)->Vec<u8>{
-		self.0
-	}
-}
-#[derive(Clone)]
-pub struct VvdData(Vec<u8>);
-impl VvdData{
-	pub const fn new(value:Vec<u8>)->Self{
-		Self(value)
-	}
-	pub fn get(self)->Vec<u8>{
-		self.0
-	}
-}
-
-#[derive(Clone)]
-pub struct ModelData{
-	pub mdl:MdlData,
-	pub vtx:VtxData,
-	pub vvd:VvdData,
-}
-
-//meshes is more prone to failure
-pub struct Meshes{
-	meshes:Vec<Option<ModelData>>,
-}
-impl Meshes{
-	pub(crate) const fn new(meshes:Vec<Option<ModelData>>)->Self{
-		Self{
-			meshes,
-		}
-	}
-	pub fn get_texture(&self,texture_id:MeshId)->Option<&ModelData>{
-		self.meshes.get(texture_id.get() as usize)?.as_ref()
-	}
-	pub fn into_iter(self)->impl Iterator<Item=(MeshId,ModelData)>{
-		self.meshes.into_iter().enumerate().filter_map(|(mesh_id,maybe_mesh)|
-			maybe_mesh.map(|mesh|(MeshId::new(mesh_id as u32),mesh))
-		)
-	}
-}

lib/fixed_wide/Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 name = "fixed_wide"
-version = "0.1.1"
-edition = "2021"
+version = "0.3.0"
+edition = "2024"
 repository = "https://git.itzana.me/StrafesNET/strafe-project"
 license = "MIT OR Apache-2.0"
 description = "Fixed point numbers with optional widening Mul operator."
@@ -14,7 +14,10 @@ wide-mul=[]
 zeroes=["dep:arrayvec"]
 
 [dependencies]
-bnum = "0.12.0"
+bnum = "0.14.3"
 arrayvec = { version = "0.7.6", optional = true }
 paste = "1.0.15"
-ratio_ops = { path = "../ratio_ops", registry = "strafesnet", optional = true }
+ratio_ops = { workspace = true, optional = true }
+
+[lints]
+workspace = true

lib/fixed_wide/src/tests.rs

@@ -1,195 +1,273 @@
-use crate::types::I32F32;
-use crate::types::I256F256;
+use crate::fixed::Fixed;
+use crate::types::{F64_32,F128_64,F192_96,F512_256};
 
 #[test]
 fn you_can_add_numbers(){
-	let a=I256F256::from((3i128*2).pow(4));
-	assert_eq!(a+a,I256F256::from((3i128*2).pow(4)*2));
+	let a=F512_256::from((3i128*2).pow(4));
+	assert_eq!(a+a,F512_256::from((3i128*2).pow(4)*2));
+}
+
+macro_rules! test_bit_by_bit{
+	($n:expr,$float:ty,$mantissa_bits:expr)=>{{
+		const MANT:u64=(1<<$mantissa_bits)-1;
+		// all bits in range
+		for i in 0..$n-$mantissa_bits{
+			let a=Fixed::<{$n/8},{$n>>1}>::from_bits(bnum::cast::As::as_::<bnum::Int::<{$n/8}>>(MANT).shl(i));
+			let b=(MANT as $float)*(2.0 as $float).powi(i as i32-{$n>>1});
+			let f:$float=a.into();
+			assert_eq!(f,b,"F{}_{} Into float {i}",$n,$n>>1);
+			assert_eq!(a,b.try_into().unwrap(),"F{}_{} From float {i}",$n,$n>>1);
+		}
+		// underflow
+		for i in 0u32..$mantissa_bits{
+			let a=Fixed::<{$n/8},{$n>>1}>::from_bits(bnum::cast::As::as_::<bnum::Int::<{$n/8}>>(MANT>>i));
+			let b=((MANT>>i) as $float)*(2.0 as $float).powi(-{$n>>1});
+			let f:$float=a.into();
+			assert_eq!(f,b,"Underflow F{}_{} Into float {i}",$n,$n>>1);
+			assert_eq!(a,b.try_into().unwrap(),"Underflow F{}_{} From float {i}",$n,$n>>1);
+		}
+	}};
+}
+
+#[test]
+fn test_many(){
+	test_bit_by_bit!(64,f32,24);
+	test_bit_by_bit!(128,f32,24);
+	// f32 is reaching its limits here
+	// test_bit_by_bit!(256,f32,24);
+	// test_bit_by_bit!(512,f32,24);
+	test_bit_by_bit!(64,f64,53);
+	test_bit_by_bit!(128,f64,53);
+	test_bit_by_bit!(256,f64,53);
+	test_bit_by_bit!(512,f64,53);
 }
 
 #[test]
 fn to_f32(){
-	let a=I256F256::from(1)>>2;
+	let a=F64_32::ZERO;
+	let f:f32=a.into();
+	assert_eq!(f,0.0f32);
+	let a=F64_32::from(1)>>2;
 	let f:f32=a.into();
 	assert_eq!(f,0.25f32);
 	let f:f32=(-a).into();
 	assert_eq!(f,-0.25f32);
-	let a=I256F256::from(0);
+	let a=F64_32::MIN;
+	let f:f32=a.into();
+	assert_eq!(f,i32::MIN as f32);
+	let a=F512_256::from(1)>>2;
+	let f:f32=a.into();
+	assert_eq!(f,0.25f32);
+	let f:f32=(-a).into();
+	assert_eq!(f,-0.25f32);
+	let a=F512_256::from(0);
 	let f:f32=(-a).into();
 	assert_eq!(f,0f32);
-	let a=I256F256::from(237946589723468975i64)<<16;
+	let a=F512_256::from(237946589723468975i64)<<16;
 	let f:f32=a.into();
 	assert_eq!(f,237946589723468975f32*2.0f32.powi(16));
 }
 
 #[test]
 fn to_f64(){
-	let a=I256F256::from(1)>>2;
+	let a=F64_32::ZERO;
+	let f:f64=a.into();
+	assert_eq!(f,0.0f64);
+	let a=F64_32::from(1)>>2;
 	let f:f64=a.into();
 	assert_eq!(f,0.25f64);
 	let f:f64=(-a).into();
 	assert_eq!(f,-0.25f64);
-	let a=I256F256::from(0);
+	let a=F64_32::MIN;
+	let f:f64=a.into();
+	assert_eq!(f,i32::MIN as f64);
+	let a=F512_256::from(1)>>2;
+	let f:f64=a.into();
+	assert_eq!(f,0.25f64);
+	let f:f64=(-a).into();
+	assert_eq!(f,-0.25f64);
+	let a=F512_256::from(0);
 	let f:f64=(-a).into();
 	assert_eq!(f,0f64);
-	let a=I256F256::from(237946589723468975i64)<<16;
+	let a=F512_256::from(237946589723468975i64)<<16;
 	let f:f64=a.into();
 	assert_eq!(f,237946589723468975f64*2.0f64.powi(16));
 }
 
 #[test]
 fn from_f32(){
-	let a=I256F256::from(1)>>2;
-	let b:Result<I256F256,_>=0.25f32.try_into();
+	let a=F64_32::ZERO;
+	let b:Result<F64_32,_>=0.0f32.try_into();
 	assert_eq!(b,Ok(a));
-	let a=I256F256::from(-1)>>2;
-	let b:Result<I256F256,_>=(-0.25f32).try_into();
+	let a=F512_256::from(1)>>2;
+	let b:Result<F512_256,_>=0.25f32.try_into();
 	assert_eq!(b,Ok(a));
-	let a=I256F256::from(0);
-	let b:Result<I256F256,_>=0.try_into();
+	let a=F512_256::from(-1)>>2;
+	let b:Result<F512_256,_>=(-0.25f32).try_into();
 	assert_eq!(b,Ok(a));
-	let a=I256F256::from(0b101011110101001010101010000000000000000000000000000i64)<<16;
-	let b:Result<I256F256,_>=(0b101011110101001010101010000000000000000000000000000u64 as f32*2.0f32.powi(16)).try_into();
+	let a=F512_256::from(0);
+	let b:Result<F512_256,_>=0.try_into();
+	assert_eq!(b,Ok(a));
+	let a=F512_256::from(0b101011110101001010101010000000000000000000000000000i64)<<16;
+	let b:Result<F512_256,_>=(0b101011110101001010101010000000000000000000000000000u64 as f32*2.0f32.powi(16)).try_into();
 	assert_eq!(b,Ok(a));
 	//I32F32::MAX into f32 is truncated into this value
-	let a=I32F32::raw(0b111111111111111111111111000000000000000000000000000000000000000i64);
-	let b:Result<I32F32,_>=Into::<f32>::into(I32F32::MAX).try_into();
+	let a=F64_32::raw(0b111111111111111111111111000000000000000000000000000000000000000i64);
+	let b:Result<F64_32,_>=Into::<f32>::into(F64_32::MAX).try_into();
 	assert_eq!(b,Ok(a));
 	//I32F32::MIN hits a special case since it's not representable as a positive signed integer
 	//TODO: don't return an overflow because this is technically possible
-	let a=I32F32::MIN;
-	let b:Result<I32F32,_>=Into::<f32>::into(I32F32::MIN).try_into();
+	let a=F64_32::MIN;
+	let f:f32=a.into();
+	let b:Result<F64_32,_>=f.try_into();
 	assert_eq!(b,Err(crate::fixed::FixedFromFloatError::Overflow));
 	//16 is within the 24 bits of float precision
-	let b:Result<I32F32,_>=Into::<f32>::into(-I32F32::MIN.fix_2()).try_into();
+	let a=-F64_32::MIN.widen_128();
+	let f:f32=a.into();
+	let b:Result<F64_32,_>=f.try_into();
 	assert_eq!(b,Err(crate::fixed::FixedFromFloatError::Overflow));
-	let b:Result<I32F32,_>=f32::MIN_POSITIVE.try_into();
+	let b:Result<F64_32,_>=f32::MIN_POSITIVE.try_into();
 	assert_eq!(b,Err(crate::fixed::FixedFromFloatError::Underflow));
 	//test many cases
 	for i in 0..64{
-		let a=crate::fixed::Fixed::<2,64>::raw_digit(0b111111111111111111111111000000000000000000000000000000000000000i64)<<i;
+		let a=F128_64::from_u64(0b111111111111111111111111000000000000000000000000000000000000000u64)<<i;
 		let f:f32=a.into();
-		let b:Result<crate::fixed::Fixed<2,64>,_>=f.try_into();
+		let b:Result<F128_64,_>=f.try_into();
 		assert_eq!(b,Ok(a));
 	}
 }
 
 #[test]
 fn from_f64(){
-	let a=I256F256::from(1)>>2;
-	let b:Result<I256F256,_>=0.25f64.try_into();
+	let a=F64_32::ZERO;
+	let b:Result<F64_32,_>=0.0f64.try_into();
 	assert_eq!(b,Ok(a));
-	let a=I256F256::from(-1)>>2;
-	let b:Result<I256F256,_>=(-0.25f64).try_into();
+	let a=F512_256::from(1)>>2;
+	let b:Result<F512_256,_>=0.25f64.try_into();
 	assert_eq!(b,Ok(a));
-	let a=I256F256::from(0);
-	let b:Result<I256F256,_>=0.try_into();
+	let a=F512_256::from(-1)>>2;
+	let b:Result<F512_256,_>=(-0.25f64).try_into();
 	assert_eq!(b,Ok(a));
-	let a=I256F256::from(0b101011110101001010101010000000000000000000000000000i64)<<16;
-	let b:Result<I256F256,_>=(0b101011110101001010101010000000000000000000000000000u64 as f64*2.0f64.powi(16)).try_into();
+	let a=F512_256::from(0);
+	let b:Result<F512_256,_>=0.try_into();
+	assert_eq!(b,Ok(a));
+	let a=F512_256::from(0b101011110101001010101010000000000000000000000000000i64)<<16;
+	let b:Result<F512_256,_>=(0b101011110101001010101010000000000000000000000000000u64 as f64*2.0f64.powi(16)).try_into();
 	assert_eq!(b,Ok(a));
 }
 
 #[test]
 fn you_can_shr_numbers(){
-	let a=I32F32::from(4);
-	assert_eq!(a>>1,I32F32::from(2));
+	let a=F64_32::from(4);
+	assert_eq!(a>>1,F64_32::from(2));
 }
 
 #[test]
 fn test_wide_mul(){
-	let a=I32F32::ONE;
-	let aa=a.wide_mul_1_1(a);
-	assert_eq!(aa,crate::types::I64F64::ONE);
+	let a=F64_32::ONE;
+	let aa=a.wide_mul_64_64(a);
+	assert_eq!(aa,F128_64::ONE);
 }
 
 #[test]
 fn test_wide_div(){
-	let a=I32F32::ONE*4;
-	let b=I32F32::ONE*2;
-	let wide_a=a.wide_mul_1_1(I32F32::ONE);
-	let wide_b=b.wide_mul_1_1(I32F32::ONE);
-	let ab=a.wide_div_1_1(b);
-	assert_eq!(ab,crate::types::I64F64::ONE*2);
-	let wab=wide_a.wide_div_2_1(b);
-	assert_eq!(wab,crate::fixed::Fixed::<3,96>::ONE*2);
-	let awb=a.wide_div_1_2(wide_b);
-	assert_eq!(awb,crate::fixed::Fixed::<3,96>::ONE*2);
+	let a=F64_32::ONE*4;
+	let b=F64_32::ONE*2;
+	let wide_a=a.wide_mul_64_64(F64_32::ONE);
+	let wide_b=b.wide_mul_64_64(F64_32::ONE);
+	let ab=a.wide_div_64_64(b);
+	assert_eq!(ab,F128_64::ONE*2);
+	let wab=wide_a.wide_div_128_64(b);
+	assert_eq!(wab,F192_96::ONE*2);
+	let awb=a.wide_div_64_128(wide_b);
+	assert_eq!(awb,F192_96::ONE*2);
 }
 
 #[test]
 fn test_wide_mul_repeated() {
-	let a=I32F32::from(2);
-	let b=I32F32::from(3);
+	let a=F64_32::from(2);
+	let b=F64_32::from(3);
 
-	let w1=a.wide_mul_1_1(b);
-	let w2=w1.wide_mul_2_2(w1);
-	let w3=w2.wide_mul_4_4(w2);
+	let w1=a.wide_mul_64_64(b);
+	let w2=w1.wide_mul_128_128(w1);
+	let w3=w2.wide_mul_256_256(w2);
 
-	assert_eq!(w3,I256F256::from((3i128*2).pow(4)));
+	assert_eq!(w3,F512_256::from((3i128*2).pow(4)));
 }
 
 #[test]
 fn test_bint(){
-	let a=I32F32::ONE;
-	assert_eq!(a*2,I32F32::from(2));
+	let a=F64_32::ONE;
+	assert_eq!(a*2,F64_32::from(2));
 }
 
 #[test]
-fn test_fix(){
-	assert_eq!(I32F32::ONE.fix_8(),I256F256::ONE);
-	assert_eq!(I32F32::ONE,I256F256::ONE.fix_1());
-	assert_eq!(I32F32::NEG_ONE.fix_8(),I256F256::NEG_ONE);
-	assert_eq!(I32F32::NEG_ONE,I256F256::NEG_ONE.fix_1());
+fn test_wrap(){
+	assert_eq!(F64_32::ONE,F512_256::ONE.wrap_64());
+	assert_eq!(F64_32::NEG_ONE,F512_256::NEG_ONE.wrap_64());
+}
+#[test]
+fn test_narrow(){
+	assert_eq!(Ok(F64_32::ONE),F512_256::ONE.narrow_64());
+	assert_eq!(Ok(F64_32::NEG_ONE),F512_256::NEG_ONE.narrow_64());
+}
+#[test]
+fn test_widen(){
+	assert_eq!(F64_32::ONE.widen_512(),F512_256::ONE);
+	assert_eq!(F64_32::NEG_ONE.widen_512(),F512_256::NEG_ONE);
+}
+#[test]
+fn test_clamp(){
+	assert_eq!(F64_32::ONE,F512_256::ONE.clamp_64());
+	assert_eq!(F64_32::NEG_ONE,F512_256::NEG_ONE.clamp_64());
 }
 #[test]
 fn test_sqrt(){
-	let a=I32F32::ONE*4;
-	assert_eq!(a.sqrt(),I32F32::from(2));
+	let a=F64_32::ONE*4;
+	assert_eq!(a.sqrt(),F64_32::from(2));
 }
 #[test]
 fn test_sqrt_zero(){
-	let a=I32F32::ZERO;
-	assert_eq!(a.sqrt(),I32F32::ZERO);
+	let a=F64_32::ZERO;
+	assert_eq!(a.sqrt(),F64_32::ZERO);
 }
 #[test]
 fn test_sqrt_low(){
-	let a=I32F32::HALF;
+	let a=F64_32::HALF;
 	let b=a.fixed_mul(a);
 	assert_eq!(b.sqrt(),a);
 }
-fn find_equiv_sqrt_via_f64(n:I32F32)->I32F32{
+fn find_equiv_sqrt_via_f64(n:F64_32)->F64_32{
 	//GIMME THEM BITS BOY
-	let &[bits]=n.to_bits().to_bits().digits();
-	let ibits=bits as i64;
+	let ibits=i64::from_le_bytes(n.to_bits().to_bytes());
 	let f=(ibits as f64)/((1u64<<32) as f64);
 	let f_ans=f.sqrt();
-	let i=(f_ans*((1u64<<32) as f64)) as i64;
-	let r=I32F32::from_bits(bnum::BInt::<1>::from(i));
+	let i=(f_ans*((1u64<<32) as f64)) as u64;
+	let r=F64_32::from_u64(i);
 	//mimic the behaviour of the algorithm,
 	//return the result if it truncates to the exact answer
-	if (r+I32F32::EPSILON).wide_mul_1_1(r+I32F32::EPSILON)==n.wide_mul_1_1(I32F32::ONE){
-		return r+I32F32::EPSILON;
+	if (r+F64_32::EPSILON).wide_mul_64_64(r+F64_32::EPSILON)==n.wide_mul_64_64(F64_32::ONE){
+		return r+F64_32::EPSILON;
 	}
-	if (r-I32F32::EPSILON).wide_mul_1_1(r-I32F32::EPSILON)==n.wide_mul_1_1(I32F32::ONE){
-		return r-I32F32::EPSILON;
+	if (r-F64_32::EPSILON).wide_mul_64_64(r-F64_32::EPSILON)==n.wide_mul_64_64(F64_32::ONE){
+		return r-F64_32::EPSILON;
 	}
 	return r;
 }
-fn test_exact(n:I32F32){
+fn test_exact(n:F64_32){
 	assert_eq!(n.sqrt(),find_equiv_sqrt_via_f64(n));
 }
 #[test]
 fn test_sqrt_exact(){
 	//43
 	for i in 0..((i64::MAX as f32).ln() as u32){
-		let n=I32F32::from_bits(bnum::BInt::<1>::from((i as f32).exp() as i64));
+		let n=F64_32::from_u64((i as f32).exp() as u64);
 		test_exact(n);
 	}
 }
 #[test]
 fn test_sqrt_max(){
-	let a=I32F32::MAX;
+	let a=F64_32::MAX;
 	test_exact(a);
 }
 #[test]
@@ -197,9 +275,9 @@ fn test_sqrt_max(){
 fn test_zeroes_normal(){
 	// (x-1)*(x+1)
 	// x^2-1
-	let zeroes=I32F32::zeroes2(I32F32::NEG_ONE,I32F32::ZERO,I32F32::ONE);
+	let zeroes=F64_32::zeroes2(F64_32::NEG_ONE,F64_32::ZERO,F64_32::ONE);
 	assert_eq!(zeroes,arrayvec::ArrayVec::from_iter([I32F32::NEG_ONE,I32F32::ONE]));
-	let zeroes=I32F32::zeroes2(I32F32::NEG_ONE*3,I32F32::ONE*2,I32F32::ONE);
+	let zeroes=F64_32::zeroes2(F64_32::NEG_ONE*3,F64_32::ONE*2,F64_32::ONE);
 	assert_eq!(zeroes,arrayvec::ArrayVec::from_iter([I32F32::NEG_ONE*3,I32F32::ONE]));
 }
 #[test]
@@ -207,12 +285,25 @@ fn test_zeroes_normal(){
 fn test_zeroes_deferred_division(){
 	// (x-1)*(x+1)
 	// x^2-1
-	let zeroes=I32F32::zeroes2(I32F32::NEG_ONE,I32F32::ZERO,I32F32::ONE);
+	let zeroes=F64_32::zeroes2(F64_32::NEG_ONE,F64_32::ZERO,F64_32::ONE);
 	assert_eq!(
 		zeroes,
 		arrayvec::ArrayVec::from_iter([
-			ratio_ops::ratio::Ratio::new(I32F32::ONE*2,I32F32::NEG_ONE*2),
-			ratio_ops::ratio::Ratio::new(I32F32::ONE*2,I32F32::ONE*2),
+			ratio_ops::ratio::Ratio::new(F64_32::ONE*2,F64_32::NEG_ONE*2),
+			ratio_ops::ratio::Ratio::new(F64_32::ONE*2,F64_32::ONE*2),
 		])
 	);
 }
+
+#[test]
+fn test_debug(){
+	assert_eq!(format!("{:?}",F64_32::EPSILON),"0.00000001");
+	assert_eq!(format!("{:?}",F64_32::ONE),"1.00000000");
+	assert_eq!(format!("{:?}",F64_32::TWO),"2.00000000");
+	assert_eq!(format!("{:?}",F64_32::MAX),"7fffffff.ffffffff");
+	assert_eq!(format!("{:?}",F64_32::try_from(core::f64::consts::PI).unwrap()),"3.243f6a88");
+	assert_eq!(format!("{:?}",F64_32::NEG_EPSILON),"-0.00000001");
+	assert_eq!(format!("{:?}",F64_32::NEG_ONE),"-1.00000000");
+	assert_eq!(format!("{:?}",F64_32::NEG_TWO),"-2.00000000");
+	assert_eq!(format!("{:?}",F64_32::MIN),"-80000000.00000000");
+}

lib/fixed_wide/src/types.rs

@@ -1,4 +1,7 @@
-pub type I32F32=crate::fixed::Fixed<1,32>;
-pub type I64F64=crate::fixed::Fixed<2,64>;
-pub type I128F128=crate::fixed::Fixed<4,128>;
-pub type I256F256=crate::fixed::Fixed<8,256>;
+use crate::fixed::BNUM_DIGIT_WIDTH;
+pub type F64_32=crate::fixed::Fixed<{64/BNUM_DIGIT_WIDTH},32>;
+pub type F128_64=crate::fixed::Fixed<{128/BNUM_DIGIT_WIDTH},64>;
+pub type F192_96=crate::fixed::Fixed<{192/BNUM_DIGIT_WIDTH},96>;
+pub type F256_128=crate::fixed::Fixed<{256/BNUM_DIGIT_WIDTH},128>;
+pub type F320_160=crate::fixed::Fixed<{320/BNUM_DIGIT_WIDTH},160>;
+pub type F512_256=crate::fixed::Fixed<{512/BNUM_DIGIT_WIDTH},256>;

lib/fixed_wide/src/zeroes.rs

@@ -1,39 +1,42 @@
 use crate::fixed::Fixed;
+use crate::fixed::BNUM_DIGIT_WIDTH;
 
 use arrayvec::ArrayVec;
 use std::cmp::Ordering;
 macro_rules! impl_zeroes{
 	($n:expr)=>{
-		impl Fixed<$n,{$n*32}>{
+		impl Fixed<{$n/BNUM_DIGIT_WIDTH},{$n>>1}>{
 			#[inline]
 			pub fn zeroes2(a0:Self,a1:Self,a2:Self)->ArrayVec<<Self as core::ops::Div>::Output,2>{
 				let a2pos=match a2.cmp(&Self::ZERO){
 					Ordering::Greater=>true,
-					Ordering::Equal=>return ArrayVec::from_iter(Self::zeroes1(a0,a1).into_iter()),
+					Ordering::Equal=>return ArrayVec::from_iter(Self::zeroes1(a0,a1)),
 					Ordering::Less=>false,
 				};
-				let radicand=a1*a1-a2*a0*4;
+				let radicand=a1*a1-((a2*a0)<<2);
 				match radicand.cmp(&<Self as core::ops::Mul>::Output::ZERO){
 					Ordering::Greater=>{
+						// using wrap because sqrt always halves the number of leading digits.
+						// clamp would be more defensive, but is slower.
 						paste::item!{
-						let planar_radicand=radicand.sqrt().[<fix_ $n>]();
+						let planar_radicand=radicand.sqrt().[<wrap_ $n>]();
 						}
 						//sort roots ascending and avoid taking the difference of large numbers
-						let zeroes=match (a2pos,Self::ZERO<a1){
-							(true, true )=>[(-a1-planar_radicand)/(a2*2),(a0*2)/(-a1-planar_radicand)],
-							(true, false)=>[(a0*2)/(-a1+planar_radicand),(-a1+planar_radicand)/(a2*2)],
-							(false,true )=>[(a0*2)/(-a1-planar_radicand),(-a1-planar_radicand)/(a2*2)],
-							(false,false)=>[(-a1+planar_radicand)/(a2*2),(a0*2)/(-a1+planar_radicand)],
+						let zeroes=match (a2pos,a1.is_positive()){
+							(true, true )=>[(-a1-planar_radicand)/(a2<<1),(a0<<1)/(-a1-planar_radicand)],
+							(true, false)=>[(a0<<1)/(-a1+planar_radicand),(-a1+planar_radicand)/(a2<<1)],
+							(false,true )=>[(a0<<1)/(-a1-planar_radicand),(-a1-planar_radicand)/(a2<<1)],
+							(false,false)=>[(-a1+planar_radicand)/(a2<<1),(a0<<1)/(-a1+planar_radicand)],
 						};
 						ArrayVec::from_iter(zeroes)
 					},
-					Ordering::Equal=>ArrayVec::from_iter([(a1)/(a2*-2)]),
+					Ordering::Equal=>ArrayVec::from_iter([(a1)/(-a2<<1)]),
 					Ordering::Less=>ArrayVec::new_const(),
 				}
 			}
 			#[inline]
 			pub fn zeroes1(a0:Self,a1:Self)->ArrayVec<<Self as core::ops::Div>::Output,1>{
-				if a1==Self::ZERO{
+				if a1.is_zero(){
 					ArrayVec::new_const()
 				}else{
 					ArrayVec::from_iter([(-a0)/(a1)])
@@ -42,10 +45,10 @@ macro_rules! impl_zeroes{
 		}
 	};
 }
-impl_zeroes!(1);
-impl_zeroes!(2);
-impl_zeroes!(3);
-impl_zeroes!(4);
+impl_zeroes!(64);
+impl_zeroes!(128);
+impl_zeroes!(192);
+impl_zeroes!(256);
 //sqrt doubles twice!
 //impl_zeroes!(5);
 //impl_zeroes!(6);

lib/linear_ops/Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 name = "linear_ops"
-version = "0.1.0"
-edition = "2021"
+version = "0.2.0"
+edition = "2024"
 repository = "https://git.itzana.me/StrafesNET/strafe-project"
 license = "MIT OR Apache-2.0"
 description = "Vector/Matrix operations using trait bounds."
@@ -14,9 +14,12 @@ fixed-wide=["dep:fixed_wide","dep:paste"]
 deferred-division=["dep:ratio_ops"]
 
 [dependencies]
-ratio_ops = { path = "../ratio_ops", registry = "strafesnet", optional = true }
-fixed_wide = { path = "../fixed_wide", registry = "strafesnet", optional = true }
+ratio_ops = { workspace = true, optional = true }
+fixed_wide = { workspace = true, optional = true }
 paste = { version = "1.0.15", optional = true }
 
 [dev-dependencies]
-fixed_wide = { path = "../fixed_wide", registry = "strafesnet", features = ["wide-mul"] }
+fixed_wide = { workspace = true, features = ["wide-mul"] }
+
+[lints]
+workspace = true

lib/linear_ops/src/macros/fixed_wide.rs

@@ -5,17 +5,17 @@ macro_rules! impl_fixed_wide_vector_not_const_generic {
 		(),
 		$n:expr
 	) => {
-		impl<const N:usize> Vector<N,fixed_wide::fixed::Fixed<$n,{$n*32}>>{
+		impl<const N:usize> Vector<N,fixed_wide::fixed::Fixed<{$n>>3},{$n>>1}>>{
 			#[inline]
-			pub fn length(self)-><fixed_wide::fixed::Fixed::<$n,{$n*32}> as core::ops::Mul>::Output{
+			pub fn length(self)-><fixed_wide::fixed::Fixed::<{$n>>3},{$n>>1}> as core::ops::Mul>::Output{
 				self.length_squared().sqrt_unchecked()
 			}
 			#[inline]
-			pub fn with_length<U,V>(self,length:U)-><Vector<N,V> as core::ops::Div<<fixed_wide::fixed::Fixed::<$n,{$n*32}> as core::ops::Mul>::Output>>::Output
+			pub fn with_length<U,V>(self,length:U)-><Vector<N,V> as core::ops::Div<<fixed_wide::fixed::Fixed::<{$n>>3},{$n>>1}> as core::ops::Mul>::Output>>::Output
 				where
-					fixed_wide::fixed::Fixed<$n,{$n*32}>:core::ops::Mul<U,Output=V>,
+					fixed_wide::fixed::Fixed<{$n>>3},{$n>>1}>:core::ops::Mul<U,Output=V>,
 					U:Copy,
-					V:core::ops::Div<<fixed_wide::fixed::Fixed::<$n,{$n*32}> as core::ops::Mul>::Output>,
+					V:core::ops::Div<<fixed_wide::fixed::Fixed::<{$n>>3},{$n>>1}> as core::ops::Mul>::Output>,
 			{
 				self*length/self.length()
 			}
@@ -27,7 +27,7 @@ macro_rules! impl_fixed_wide_vector_not_const_generic {
 #[macro_export(local_inner_macros)]
 macro_rules! macro_4 {
 	( $macro: ident, $any:tt ) => {
-		$crate::macro_repeated!($macro,$any,1,2,3,4);
+		$crate::macro_repeated!($macro,$any,64,128,192,256);
 	}
 }
 
@@ -38,40 +38,95 @@ macro_rules! impl_fixed_wide_vector {
 		$crate::macro_4!(impl_fixed_wide_vector_not_const_generic,());
 		// I LOVE NOT BEING ABLE TO USE CONST GENERICS
 		$crate::macro_repeated!(
-			impl_fix_not_const_generic,(),
-			(1,1),(2,1),(3,1),(4,1),(5,1),(6,1),(7,1),(8,1),(9,1),(10,1),(11,1),(12,1),(13,1),(14,1),(15,1),(16,1),
-			(1,2),(2,2),(3,2),(4,2),(5,2),(6,2),(7,2),(8,2),(9,2),(10,2),(11,2),(12,2),(13,2),(14,2),(15,2),(16,2),
-			(1,3),(2,3),(3,3),(4,3),(5,3),(6,3),(7,3),(8,3),(9,3),(10,3),(11,3),(12,3),(13,3),(14,3),(15,3),(16,3),
-			(1,4),(2,4),(3,4),(4,4),(5,4),(6,4),(7,4),(8,4),(9,4),(10,4),(11,4),(12,4),(13,4),(14,4),(15,4),(16,4),
-			(1,5),(2,5),(3,5),(4,5),(5,5),(6,5),(7,5),(8,5),(9,5),(10,5),(11,5),(12,5),(13,5),(14,5),(15,5),(16,5),
-			(1,6),(2,6),(3,6),(4,6),(5,6),(6,6),(7,6),(8,6),(9,6),(10,6),(11,6),(12,6),(13,6),(14,6),(15,6),(16,6),
-			(1,7),(2,7),(3,7),(4,7),(5,7),(6,7),(7,7),(8,7),(9,7),(10,7),(11,7),(12,7),(13,7),(14,7),(15,7),(16,7),
-			(1,8),(2,8),(3,8),(4,8),(5,8),(6,8),(7,8),(8,8),(9,8),(10,8),(11,8),(12,8),(13,8),(14,8),(15,8),(16,8),
-			(1,9),(2,9),(3,9),(4,9),(5,9),(6,9),(7,9),(8,9),(9,9),(10,9),(11,9),(12,9),(13,9),(14,9),(15,9),(16,9),
-			(1,10),(2,10),(3,10),(4,10),(5,10),(6,10),(7,10),(8,10),(9,10),(10,10),(11,10),(12,10),(13,10),(14,10),(15,10),(16,10),
-			(1,11),(2,11),(3,11),(4,11),(5,11),(6,11),(7,11),(8,11),(9,11),(10,11),(11,11),(12,11),(13,11),(14,11),(15,11),(16,11),
-			(1,12),(2,12),(3,12),(4,12),(5,12),(6,12),(7,12),(8,12),(9,12),(10,12),(11,12),(12,12),(13,12),(14,12),(15,12),(16,12),
-			(1,13),(2,13),(3,13),(4,13),(5,13),(6,13),(7,13),(8,13),(9,13),(10,13),(11,13),(12,13),(13,13),(14,13),(15,13),(16,13),
-			(1,14),(2,14),(3,14),(4,14),(5,14),(6,14),(7,14),(8,14),(9,14),(10,14),(11,14),(12,14),(13,14),(14,14),(15,14),(16,14),
-			(1,15),(2,15),(3,15),(4,15),(5,15),(6,15),(7,15),(8,15),(9,15),(10,15),(11,15),(12,15),(13,15),(14,15),(15,15),(16,15),
-			(1,16),(2,16),(3,16),(4,16),(5,16),(6,16),(7,16),(8,16),(9,16),(10,16),(11,16),(12,16),(13,16),(14,16),(15,16),(16,16)
+			impl_narrow_not_const_generic,(),
+			(128,64),(192,64),(256,64),(320,64),(384,64),(448,64),(512,64),(576,64),(640,64),(704,64),(768,64),(832,64),(896,64),(960,64),(1024,64),(1088,64),
+			(192,128),(256,128),(320,128),(384,128),(448,128),(512,128),(576,128),(640,128),(704,128),(768,128),(832,128),(896,128),(960,128),(1024,128),
+			(256,192),(320,192),(384,192),(448,192),(512,192),(576,192),(640,192),(704,192),(768,192),(832,192),(896,192),(960,192),(1024,192),
+			(320,256),(384,256),(448,256),(512,256),(576,256),(640,256),(704,256),(768,256),(832,256),(896,256),(960,256),(1024,256),
+			(384,320),(448,320),(512,320),(576,320),(640,320),(704,320),(768,320),(832,320),(896,320),(960,320),(1024,320),
+			(448,384),(512,384),(576,384),(640,384),(704,384),(768,384),(832,384),(896,384),(960,384),(1024,384),
+			(512,448),(576,448),(640,448),(704,448),(768,448),(832,448),(896,448),(960,448),(1024,448),
+			(576,512),(640,512),(704,512),(768,512),(832,512),(896,512),(960,512),(1024,512),
+			(640,576),(704,576),(768,576),(832,576),(896,576),(960,576),(1024,576),
+			(704,640),(768,640),(832,640),(896,640),(960,640),(1024,640),
+			(768,704),(832,704),(896,704),(960,704),(1024,704),
+			(832,768),(896,768),(960,768),(1024,768),
+			(896,832),(960,832),(1024,832),
+			(960,896),(1024,896),
+			(1024,960)
 		);
+		$crate::macro_repeated!(
+			impl_widen_not_const_generic,(),
+			(64,128),
+			(64,192),(128,192),
+			(64,256),(128,256),(192,256),
+			(64,320),(128,320),(192,320),(256,320),
+			(64,384),(128,384),(192,384),(256,384),(320,384),
+			(64,448),(128,448),(192,448),(256,448),(320,448),(384,448),
+			(64,512),(128,512),(192,512),(256,512),(320,512),(384,512),(448,512),
+			(64,576),(128,576),(192,576),(256,576),(320,576),(384,576),(448,576),(512,576),
+			(64,640),(128,640),(192,640),(256,640),(320,640),(384,640),(448,640),(512,640),(576,640),
+			(64,704),(128,704),(192,704),(256,704),(320,704),(384,704),(448,704),(512,704),(576,704),(640,704),
+			(64,768),(128,768),(192,768),(256,768),(320,768),(384,768),(448,768),(512,768),(576,768),(640,768),(704,768),
+			(64,832),(128,832),(192,832),(256,832),(320,832),(384,832),(448,832),(512,832),(576,832),(640,832),(704,832),(768,832),
+			(64,896),(128,896),(192,896),(256,896),(320,896),(384,896),(448,896),(512,896),(576,896),(640,896),(704,896),(768,896),(832,896),
+			(64,960),(128,960),(192,960),(256,960),(320,960),(384,960),(448,960),(512,960),(576,960),(640,960),(704,960),(768,960),(832,960),(896,960),
+			(64,1024),(128,1024),(192,1024),(256,1024),(320,1024),(384,1024),(448,1024),(512,1024),(576,1024),(640,1024),(704,1024),(768,1024),(832,1024),(896,1024),(960,1024),
+			(64,1088)
+		);
+		impl<const N:usize,T:fixed_wide::fixed::Wrap<U>,U> fixed_wide::fixed::Wrap<Vector<N,U>> for Vector<N,T>
+		{
+			#[inline]
+			fn wrap(self)->Vector<N,U>{
+				self.map(|t|t.wrap())
+			}
+		}
+		impl<const N:usize,T:fixed_wide::fixed::Clamp<U>,U> fixed_wide::fixed::Clamp<Vector<N,U>> for Vector<N,T>
+		{
+			#[inline]
+			fn clamp(self)->Vector<N,U>{
+				self.map(|t|t.clamp())
+			}
+		}
 	};
 }
-
 #[doc(hidden)]
 #[macro_export(local_inner_macros)]
-macro_rules! impl_fix_not_const_generic{
+macro_rules! impl_narrow_not_const_generic{
 	(
 		(),
 		($lhs:expr,$rhs:expr)
 	)=>{
-		impl<const N:usize> Vector<N,fixed_wide::fixed::Fixed<$lhs,{$lhs*32}>>
-		{
-			paste::item!{
+		paste::item!{
+			impl<const N:usize> Vector<N,fixed_wide::fixed::Fixed<{$lhs>>3},{$lhs>>1}>>{
 				#[inline]
-				pub fn [<fix_ $rhs>](self)->Vector<N,fixed_wide::fixed::Fixed<$rhs,{$rhs*32}>>{
-					self.map(|t|t.[<fix_ $rhs>]())
+				pub fn [<wrap_ $rhs>](self)->Vector<N,fixed_wide::fixed::Fixed<{$rhs>>3},{$rhs>>1}>>{
+					self.map(|t|t.[<wrap_ $rhs>]())
+				}
+				#[inline]
+				pub fn [<narrow_ $rhs>](self)->Vector<N,Result<fixed_wide::fixed::Fixed<{$rhs>>3},{$rhs>>1}>,fixed_wide::fixed::NarrowError>>{
+					self.map(|t|t.[<narrow_ $rhs>]())
+				}
+				#[inline]
+				pub fn [<clamp_ $rhs>](self)->Vector<N,fixed_wide::fixed::Fixed<{$rhs>>3},{$rhs>>1}>>{
+					self.map(|t|t.[<clamp_ $rhs>]())
+				}
+			}
+		}
+	}
+}
+#[doc(hidden)]
+#[macro_export(local_inner_macros)]
+macro_rules! impl_widen_not_const_generic{
+	(
+		(),
+		($lhs:expr,$rhs:expr)
+	)=>{
+		paste::item!{
+			impl<const N:usize> Vector<N,fixed_wide::fixed::Fixed<{$lhs>>3},{$lhs>>1}>>{
+				#[inline]
+				pub fn [<widen_ $rhs>](self)->Vector<N,fixed_wide::fixed::Fixed<{$rhs>>3},{$rhs>>1}>>{
+					self.map(|t|t.[<widen_ $rhs>]())
 				}
 			}
 		}

lib/linear_ops/src/macros/matrix.rs

@@ -204,13 +204,21 @@ macro_rules! impl_matrix_named_fields_shape {
 			type Target=$struct_outer<Vector<$size_inner,T>>;
 			#[inline]
 			fn deref(&self)->&Self::Target{
-				unsafe{core::mem::transmute(&self.array)}
+				// This cast is valid because Matrix has #[repr(transparent)]
+				let ptr:*const [[T;$size_inner];$size_outer]=&self.array;
+				let ptr=ptr as *const Self::Target;
+				// SAFETY: this pointer is non-null because it comes from a reference
+				unsafe{&*ptr}
 			}
 		}
 		impl<T> core::ops::DerefMut for Matrix<$size_outer,$size_inner,T>{
 			#[inline]
 			fn deref_mut(&mut self)->&mut Self::Target{
-				unsafe{core::mem::transmute(&mut self.array)}
+				// This cast is valid because Matrix has #[repr(transparent)]
+				let ptr:*mut [[T;$size_inner];$size_outer]=&mut self.array;
+				let ptr=ptr as *mut Self::Target;
+				// SAFETY: this pointer is non-null because it comes from a reference
+				unsafe{&mut*ptr}
 			}
 		}
 	}

lib/linear_ops/src/macros/vector.rs

@@ -58,6 +58,15 @@ macro_rules! impl_vector {
 			}
 		}
 
+		impl<const N:usize,T,E:std::fmt::Debug> Vector<N,Result<T,E>>{
+			#[inline]
+			pub fn unwrap(self)->Vector<N,T>{
+				Vector{
+					array:self.array.map(Result::unwrap)
+				}
+			}
+		}
+
 		impl<const N:usize,T:Ord> Vector<N,T>{
 			#[inline]
 			pub fn min(self,rhs:Self)->Self{
@@ -321,13 +330,21 @@ macro_rules! impl_vector_named_fields {
 			type Target=$struct<T>;
 			#[inline]
 			fn deref(&self)->&Self::Target{
-				unsafe{core::mem::transmute(&self.array)}
+				// This cast is valid because Vector has #[repr(transparent)]
+				let ptr:*const [T;$size]=&self.array;
+				let ptr=ptr as *const Self::Target;
+				// SAFETY: this pointer is non-null because it comes from a reference
+				unsafe{&*ptr}
 			}
 		}
 		impl<T> core::ops::DerefMut for Vector<$size,T>{
 			#[inline]
 			fn deref_mut(&mut self)->&mut Self::Target{
-				unsafe{core::mem::transmute(&mut self.array)}
+				// This cast is valid because Vector has #[repr(transparent)]
+				let ptr:*mut [T;$size]=&mut self.array;
+				let ptr=ptr as *mut Self::Target;
+				// SAFETY: this pointer is non-null because it comes from a reference
+				unsafe{&mut*ptr}
 			}
 		}
 	}

lib/linear_ops/src/matrix.rs

@@ -1,5 +1,6 @@
 use crate::vector::Vector;
 
+#[repr(transparent)]
 #[derive(Clone,Copy,Debug,Hash,Eq,PartialEq)]
 pub struct Matrix<const X:usize,const Y:usize,T>{
 	pub(crate) array:[[T;Y];X],

lib/linear_ops/src/tests/fixed_wide.rs

@@ -1,9 +1,9 @@
 use crate::types::{Matrix3,Matrix3x2,Matrix3x4,Matrix4x2,Vector3};
 
-type Planar64=fixed_wide::types::I32F32;
-type Planar64Wide1=fixed_wide::types::I64F64;
+type Planar64=fixed_wide::types::F64_32;
+type Planar64Wide1=fixed_wide::types::F128_64;
 //type Planar64Wide2=fixed_wide::types::I128F128;
-type Planar64Wide3=fixed_wide::types::I256F256;
+type Planar64Wide3=fixed_wide::types::F512_256;
 
 #[test]
 fn wide_vec3(){
@@ -72,7 +72,7 @@ fn wide_matrix_det(){
 	]);
 	// In[2]:= Det[{{1, 2, 3}, {4, 5, 7}, {6, 8, 9}}]
 	// Out[2]= 7
-	assert_eq!(m.det(),fixed_wide::fixed::Fixed::<3,96>::from(7));
+	assert_eq!(m.det(),fixed_wide::types::F192_96::from(7));
 }
 
 #[test]

lib/linear_ops/src/vector.rs

@@ -3,6 +3,7 @@
 /// v.x += v.z;
 /// println!("v.x={}",v.x);
 
+#[repr(transparent)]
 #[derive(Clone,Copy,Debug,Hash,Eq,PartialEq)]
 pub struct Vector<const N:usize,T>{
 	pub(crate) array:[T;N],

lib/ratio_ops/Cargo.toml

@@ -1,10 +1,13 @@
 [package]
 name = "ratio_ops"
-version = "0.1.0"
-edition = "2021"
+version = "0.1.1"
+edition = "2024"
 repository = "https://git.itzana.me/StrafesNET/strafe-project"
 license = "MIT OR Apache-2.0"
 description = "Ratio operations using trait bounds for avoiding division like the plague."
 authors = ["Rhys Lloyd <krakow20@gmail.com>"]
 
 [dependencies]
+
+[lints]
+workspace = true

lib/ratio_ops/src/ratio.rs

@@ -268,30 +268,35 @@ impl<LhsNum,LhsDen,RhsNum,RhsDen,T,U> PartialEq<Ratio<RhsNum,RhsDen>> for Ratio<
 }
 impl<Num,Den> Eq for Ratio<Num,Den> where Self:PartialEq{}
 
+// Wow! These were both completely wrong!
+// Idea: use a 'signed' trait instead of parity and float the sign to the numerator.
 impl<LhsNum,LhsDen,RhsNum,RhsDen,T,U> PartialOrd<Ratio<RhsNum,RhsDen>> for Ratio<LhsNum,LhsDen>
 	where
 		LhsNum:Copy,
-		LhsDen:Copy,
+		LhsDen:Copy+Parity,
 		RhsNum:Copy,
-		RhsDen:Copy,
+		RhsDen:Copy+Parity,
 		LhsNum:core::ops::Mul<RhsDen,Output=T>,
+		LhsDen:core::ops::Mul<RhsNum,Output=T>,
 		RhsNum:core::ops::Mul<LhsDen,Output=U>,
-		T:PartialOrd<U>,
+		RhsDen:core::ops::Mul<LhsNum,Output=U>,
+		T:PartialOrd<U>+Ord,
 {
 	#[inline]
-	fn partial_cmp(&self,other:&Ratio<RhsNum,RhsDen>)->Option<core::cmp::Ordering>{
-		(self.num*other.den).partial_cmp(&(other.num*self.den))
+	fn partial_cmp(&self,&other:&Ratio<RhsNum,RhsDen>)->Option<core::cmp::Ordering>{
+		self.partial_cmp_ratio(other)
 	}
 }
 impl<Num,Den,T> Ord for Ratio<Num,Den>
 	where
 		Num:Copy,
-		Den:Copy,
+		Den:Copy+Parity,
 		Num:core::ops::Mul<Den,Output=T>,
+		Den:core::ops::Mul<Num,Output=T>,
 		T:Ord,
 {
 	#[inline]
-	fn cmp(&self,other:&Self)->std::cmp::Ordering{
-		(self.num*other.den).cmp(&(other.num*self.den))
+	fn cmp(&self,&other:&Self)->std::cmp::Ordering{
+		self.cmp_ratio(other)
 	}
 }

lib/rbx_loader/Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 name = "strafesnet_rbx_loader"
-version = "0.5.2"
-edition = "2021"
+version = "0.10.2"
+edition = "2024"
 repository = "https://git.itzana.me/StrafesNET/strafe-project"
 license = "MIT OR Apache-2.0"
 description = "Convert Roblox place and model files to StrafesNET data structures."
@@ -11,12 +11,18 @@ authors = ["Rhys Lloyd <krakow20@gmail.com>"]
 
 [dependencies]
 bytemuck = "1.14.3"
-glam = "0.29.0"
-lazy-regex = "3.1.0"
-rbx_binary = { version = "0.7.4", registry = "strafesnet" }
-rbx_dom_weak = { version = "2.7.0", registry = "strafesnet" }
-rbx_mesh = "0.1.2"
-rbx_reflection_database = { version = "0.2.10", registry = "strafesnet" }
-rbx_xml = { version = "0.13.3", registry = "strafesnet" }
-roblox_emulator = { path = "../roblox_emulator", registry = "strafesnet" }
-strafesnet_common = { path = "../common", registry = "strafesnet" }
+glam.workspace = true
+regex = { version = "1.11.3", default-features = false, features = ["unicode-perl"] }
+rbx_mesh = "0.6.0"
+rbxassetid = { version = "0.1.0", path = "../rbxassetid", registry = "strafesnet" }
+roblox_emulator = { version = "0.5.1", path = "../roblox_emulator", default-features = false, registry = "strafesnet" }
+strafesnet_common.workspace = true
+strafesnet_deferred_loader.workspace = true
+rbx_binary = "2.0.1"
+rbx_dom_weak = "4.1.0"
+rbx_reflection = "6.1.0"
+rbx_reflection_database = "2.0.2"
+rbx_xml = "2.0.1"
+
+[lints]
+workspace = true

lib/rbx_loader/src/error.rs

@@ -0,0 +1,249 @@
+use std::collections::HashSet;
+use std::num::ParseIntError;
+
+use strafesnet_common::gameplay_modes::{StageId,ModeId};
+use strafesnet_common::integer::{FixedFromFloatError,Planar64TryFromFloatError};
+
+/// A collection of errors which can be ignored at your peril
+#[derive(Debug,Default)]
+pub struct RecoverableErrors{
+	/// A basepart has an invalid / missing property.
+	pub basepart_property:Vec<InstancePath>,
+	/// A part has an unconvertable CFrame.
+	pub basepart_cframe:Vec<CFrameError>,
+	/// A part has an unconvertable Velocity.
+	pub basepart_velocity:Vec<Planar64ConvertError>,
+	/// A part has an invalid / missing property.
+	pub part_property:Vec<InstancePath>,
+	/// A part has an invalid shape.
+	pub part_shape:Vec<ShapeError>,
+	/// A meshpart has an invalid / missing property.
+	pub meshpart_property:Vec<InstancePath>,
+	/// A meshpart has no mesh.
+	pub meshpart_content:Vec<InstancePath>,
+	/// A basepart has an unsupported subclass.
+	pub unsupported_class:HashSet<String>,
+	/// A decal has an invalid / missing property.
+	pub decal_property:Vec<InstancePath>,
+	/// A SurfaceAppearance has an invalid / missing property.
+	pub surface_appearance_property:Vec<InstancePath>,
+	/// A decal has an invalid normal_id.
+	pub normal_id:Vec<NormalIdError>,
+	/// A texture has an invalid / missing property.
+	pub texture_property:Vec<InstancePath>,
+	/// A mode_id failed to parse.
+	pub mode_id_parse_int:Vec<ParseIntContext>,
+	/// There is a duplicate mode.
+	pub duplicate_mode:HashSet<ModeId>,
+	/// A mode_id failed to parse.
+	pub stage_id_parse_int:Vec<ParseIntContext>,
+	/// A Stage was duplicated leading to undefined behaviour.
+	pub duplicate_stage:HashSet<DuplicateStageError>,
+	/// A WormholeOut id failed to parse.
+	pub wormhole_out_id_parse_int:Vec<ParseIntContext>,
+	/// A WormholeOut was duplicated leading to undefined behaviour.
+	pub duplicate_wormhole_out:HashSet<u32>,
+	/// A WormholeIn id failed to parse.
+	pub wormhole_in_id_parse_int:Vec<ParseIntContext>,
+	/// A jump limit failed to parse.
+	pub jump_limit_parse_int:Vec<ParseIntContext>,
+}
+
+impl RecoverableErrors{
+	pub fn count(&self)->usize{
+		self.basepart_property.len()+
+		self.basepart_cframe.len()+
+		self.basepart_velocity.len()+
+		self.part_property.len()+
+		self.part_shape.len()+
+		self.meshpart_property.len()+
+		self.meshpart_content.len()+
+		self.unsupported_class.len()+
+		self.decal_property.len()+
+		self.surface_appearance_property.len()+
+		self.normal_id.len()+
+		self.texture_property.len()+
+		self.mode_id_parse_int.len()+
+		self.duplicate_mode.len()+
+		self.stage_id_parse_int.len()+
+		self.duplicate_stage.len()+
+		self.wormhole_out_id_parse_int.len()+
+		self.duplicate_wormhole_out.len()+
+		self.wormhole_in_id_parse_int.len()+
+		self.jump_limit_parse_int.len()
+	}
+}
+
+fn write_comma_separated<T>(
+	f:&mut std::fmt::Formatter<'_>,
+	mut it:impl Iterator<Item=T>,
+	custom_write:impl Fn(&mut std::fmt::Formatter<'_>,T)->std::fmt::Result
+)->std::fmt::Result{
+	if let Some(t)=it.next(){
+		custom_write(f,t)?;
+		for t in it{
+			write!(f,", ")?;
+			custom_write(f,t)?;
+		}
+	}
+	Ok(())
+}
+
+macro_rules! write_instance_path_error{
+	($f:ident,$self:ident,$field:ident,$class:literal,$class_plural:literal,$problem:literal)=>{
+		let len=$self.$field.len();
+		if len!=0{
+			let plural=if len==1{$class}else{$class_plural};
+			write!($f,"The following {plural} {}: ",$problem)?;
+			write_comma_separated($f,$self.$field.iter(),|f,InstancePath(path)|
+				write!(f,"{path}")
+			)?;
+			writeln!($f)?;
+		}
+	};
+}
+macro_rules! write_duplicate_error{
+	($f:ident,$self:ident,$field:ident,$class:literal,$class_plural:literal)=>{
+		let len=$self.$field.len();
+		if len!=0{
+			let plural=if len==1{$class}else{$class_plural};
+			write!($f,"The following {plural} duplicates: ")?;
+			write_comma_separated($f,$self.$field.iter(),|f,id|
+				write!(f,"{id}")
+			)?;
+			writeln!($f)?;
+		}
+	};
+}
+macro_rules! write_bespoke_error{
+	($f:ident,$self:ident,$field:ident,$class:literal,$class_plural:literal,$problem:literal,$path_field:ident,$error_field:ident)=>{
+		let len=$self.$field.len();
+		if len!=0{
+			let plural=if len==1{$class}else{$class_plural};
+			write!($f,"The following {plural} {}: ",$problem)?;
+			write_comma_separated($f,$self.$field.iter(),|f,context|
+				write!(f,"{} ({})",context.$path_field,context.$error_field)
+			)?;
+			writeln!($f)?;
+		}
+	};
+}
+
+impl core::fmt::Display for RecoverableErrors{
+	fn fmt(&self,f:&mut std::fmt::Formatter<'_>)->std::fmt::Result{
+		write_instance_path_error!(f,self,basepart_property,"BasePart is","BaseParts are","missing a property");
+		write_bespoke_error!(f,self,basepart_cframe,"BasePart","BaseParts","CFrame float convert failed",path,error);
+		write_bespoke_error!(f,self,basepart_velocity,"BasePart","BaseParts","Velocity float convert failed",path,error);
+		write_instance_path_error!(f,self,part_property,"Part is","Parts are","missing a property");
+		write_bespoke_error!(f,self,part_shape,"Part","Parts","Shape is invalid",path,shape);
+		write_instance_path_error!(f,self,meshpart_property,"MeshPart is","MeshParts are","missing a property");
+		write_instance_path_error!(f,self,meshpart_content,"MeshPart has","MeshParts have","no mesh");
+		{
+			let len=self.unsupported_class.len();
+			if len!=0{
+				let plural=if len==1{"Class is"}else{"Classes are"};
+				write!(f,"The following {plural} not supported: ")?;
+				write_comma_separated(f,self.unsupported_class.iter(),|f,classname|write!(f,"{classname}"))?;
+				writeln!(f)?;
+			}
+		}
+		write_instance_path_error!(f,self,decal_property,"Decal is","Decals are","missing a property");
+		write_instance_path_error!(f,self,surface_appearance_property,"SurfaceAppearance is","SurfaceAppearances are","missing a property");
+		write_bespoke_error!(f,self,normal_id,"Decal","Decals","NormalId is invalid",path,normal_id);
+		write_instance_path_error!(f,self,texture_property,"Texture is","Textures are","missing a property");
+		write_bespoke_error!(f,self,mode_id_parse_int,"ModeId","ModeIds","failed to parse",context,error);
+		write_duplicate_error!(f,self,duplicate_mode,"ModeId has","ModeIds have");
+		write_bespoke_error!(f,self,stage_id_parse_int,"StageId","StageIds","failed to parse",context,error);
+		write_duplicate_error!(f,self,duplicate_stage,"StageId has","StageIds have");
+		write_bespoke_error!(f,self,wormhole_out_id_parse_int,"WormholeOutId","WormholeOutIds","failed to parse",context,error);
+		write_duplicate_error!(f,self,duplicate_wormhole_out,"WormholeOutId has","WormholeOutIds have");
+		write_bespoke_error!(f,self,wormhole_in_id_parse_int,"WormholeInId","WormholeInIds","failed to parse",context,error);
+		write_bespoke_error!(f,self,jump_limit_parse_int,"jump limit","jump limits","failed to parse",context,error);
+		Ok(())
+	}
+}
+
+/// A Decal was missing required properties
+#[derive(Debug)]
+pub struct InstancePath(pub String);
+
+impl core::fmt::Display for InstancePath{
+	fn fmt(&self,f:&mut std::fmt::Formatter<'_>)->std::fmt::Result{
+		self.0.fmt(f)
+	}
+}
+
+impl InstancePath{
+	pub fn new(dom:&rbx_dom_weak::WeakDom,instance:&rbx_dom_weak::Instance)->InstancePath{
+		let mut names:Vec<_>=core::iter::successors(
+			Some(instance),
+			|i|dom.get_by_ref(i.parent())
+		).map(
+			|i|i.name.as_str()
+		).collect();
+		// discard the name of the root object
+		names.pop();
+		names.reverse();
+		InstancePath(names.join("."))
+	}
+}
+
+#[derive(Debug)]
+pub struct ParseIntContext{
+	pub context:String,
+	pub error:ParseIntError,
+}
+impl ParseIntContext{
+	pub fn parse<T:core::str::FromStr<Err=ParseIntError>>(input:&str)->Result<T,Self>{
+		input.parse().map_err(|error|ParseIntContext{
+			context:input.to_owned(),
+			error,
+		})
+	}
+}
+
+#[derive(Debug)]
+pub struct NormalIdError{
+	pub path:InstancePath,
+	pub normal_id:u32,
+}
+
+#[derive(Debug)]
+pub struct ShapeError{
+	pub path:InstancePath,
+	pub shape:u32,
+}
+
+#[derive(Debug)]
+pub enum CFrameErrorType{
+	ZeroDeterminant,
+	Convert(FixedFromFloatError),
+}
+impl core::fmt::Display for CFrameErrorType{
+	fn fmt(&self,f:&mut std::fmt::Formatter<'_>)->std::fmt::Result{
+		write!(f,"{self:?}")
+	}
+}
+
+#[derive(Debug)]
+pub struct CFrameError{
+	pub path:InstancePath,
+	pub error:CFrameErrorType,
+}
+
+#[derive(Debug)]
+pub struct Planar64ConvertError{
+	pub path:InstancePath,
+	pub error:Planar64TryFromFloatError,
+}
+
+#[derive(Debug,Hash,Eq,PartialEq)]
+pub struct DuplicateStageError{
+	pub mode_id:ModeId,
+	pub stage_id:StageId,
+}
+impl core::fmt::Display for DuplicateStageError{
+	fn fmt(&self,f:&mut std::fmt::Formatter<'_>)->std::fmt::Result{
+		write!(f,"{}-Spawn{}",self.mode_id,self.stage_id.get())
+	}
+}

lib/rbx_loader/src/lib.rs

@@ -1,35 +1,22 @@
-use std::io::Read;
 use rbx_dom_weak::WeakDom;
+use roblox_emulator::context::Context;
 
-mod rbx;
-mod mesh;
-mod primitives;
+pub use error::RecoverableErrors;
+pub use roblox_emulator::runner::Error as RunnerError;
 
-pub mod data{
-	pub struct RobloxMeshBytes(Vec<u8>);
-	impl RobloxMeshBytes{
-		pub fn new(bytes:Vec<u8>)->Self{
-			Self(bytes)
-		}
-		pub(crate) fn cursor(self)->std::io::Cursor<Vec<u8>>{
-			std::io::Cursor::new(self.0)
-		}
-	}
-}
+pub mod rbx;
+pub mod mesh;
+mod error;
+pub mod union;
+pub mod primitives;
 
 pub struct Model{
 	dom:WeakDom,
 }
 impl Model{
-	fn new(dom:WeakDom)->Self{
+	pub fn new(dom:WeakDom)->Self{
 		Self{dom}
 	}
-	pub fn into_place(self)->Place{
-		let Self{mut dom}=self;
-		let context=roblox_emulator::context::Context::from_mut(&mut dom);
-		let services=context.convert_into_place();
-		Place{dom,services}
-	}
 }
 impl AsRef<WeakDom> for Model{
 	fn as_ref(&self)->&WeakDom{
@@ -38,70 +25,39 @@ impl AsRef<WeakDom> for Model{
 }
 
 pub struct Place{
-	dom:WeakDom,
-	services:roblox_emulator::context::Services,
+	context:Context,
 }
 impl Place{
-	fn new(dom:WeakDom)->Option<Self>{
-		let context=roblox_emulator::context::Context::from_ref(&dom);
-		Some(Self{
-			services:context.find_services()?,
-			dom,
+	pub fn new(dom:WeakDom)->Result<Self,roblox_emulator::context::ServicesError>{
+		let context=Context::from_place(dom)?;
+		Ok(Self{
+			context,
 		})
 	}
-	pub fn run_scripts(&mut self){
-		let Place{dom,services}=self;
-		let runner=roblox_emulator::runner::Runner::new().unwrap();
-		let context=roblox_emulator::context::Context::from_mut(dom);
+	pub fn run_scripts(&mut self)->Result<Vec<RunnerError>,RunnerError>{
+		let Place{context}=self;
+		let runner=roblox_emulator::runner::Runner::new()?;
 		let scripts=context.scripts();
-		let runnable=runner.runnable_context_with_services(context,services).unwrap();
+		let runnable=runner.runnable_context(context)?;
+		let mut errors=Vec::new();
 		for script in scripts{
 			if let Err(e)=runnable.run_script(script){
-				println!("runner error: {e}");
+				errors.push(e);
 			}
 		}
+		Ok(errors)
 	}
 }
 impl AsRef<WeakDom> for Place{
 	fn as_ref(&self)->&WeakDom{
-		&self.dom
+		self.context.as_ref()
 	}
 }
-
-#[derive(Debug)]
-pub enum ReadError{
-	RbxBinary(rbx_binary::DecodeError),
-	RbxXml(rbx_xml::DecodeError),
-	Io(std::io::Error),
-	UnknownFileFormat,
-}
-impl std::fmt::Display for ReadError{
-	fn fmt(&self,f:&mut std::fmt::Formatter<'_>)->std::fmt::Result{
-		write!(f,"{self:?}")
+impl From<Model> for Place{
+	fn from(model:Model)->Self{
+		let context=Context::from_model(model.dom);
+		Self{
+			context,
+		}
 	}
 }
-impl std::error::Error for ReadError{}
-
-pub fn read<R:Read>(input:R)->Result<Model,ReadError>{
-	let mut buf=std::io::BufReader::new(input);
-	let peek=std::io::BufRead::fill_buf(&mut buf).map_err(ReadError::Io)?;
-	match &peek[0..8]{
-		b"<roblox!"=>rbx_binary::from_reader(buf).map(Model::new).map_err(ReadError::RbxBinary),
-		b"<roblox "=>rbx_xml::from_reader_default(buf).map(Model::new).map_err(ReadError::RbxXml),
-		_=>Err(ReadError::UnknownFileFormat),
-	}
-}
-
-//ConvertError
-
-pub fn convert<AcquireRenderConfigId,AcquireMeshId>(
-	dom:impl AsRef<WeakDom>,
-	acquire_render_config_id:AcquireRenderConfigId,
-	acquire_mesh_id:AcquireMeshId
-)->rbx::PartialMap1
-where
-	AcquireRenderConfigId:FnMut(Option<&str>)->strafesnet_common::model::RenderConfigId,
-	AcquireMeshId:FnMut(&str)->strafesnet_common::model::MeshId,
-{
-	rbx::convert(&dom.as_ref(),acquire_render_config_id,acquire_mesh_id)
-}

lib/rbx_loader/src/mesh.rs

@@ -1,12 +1,77 @@
 use std::collections::HashMap;
 
-use rbx_mesh::mesh::{Vertex2, Vertex2Truncated};
-use strafesnet_common::{integer::vec3,model::{self, ColorId, IndexedVertex, NormalId, PolygonGroup, PolygonList, PositionId, TextureCoordinateId, VertexId}};
+use rbx_mesh::mesh::{Vertex2,Vertex2Truncated};
+use strafesnet_common::aabb::Aabb;
+use strafesnet_common::integer::vec3;
+use strafesnet_common::model::{self,ColorId,IndexedVertex,PolygonGroup,PolygonList,RenderConfigId,VertexId};
+
+use crate::rbx::RobloxPartDescription;
+
+#[derive(Hash,Eq,PartialEq)]
+pub enum MeshType<'a>{
+	FileMesh,
+	Union{
+		mesh_data:&'a [u8],
+		physics_data:&'a [u8],
+		size_float_bits:[u32;3],
+		part_texture_description:RobloxPartDescription,
+	},
+}
+#[derive(Hash,Eq,PartialEq)]
+pub struct MeshIndex<'a>{
+	pub mesh_type:MeshType<'a>,
+	pub content:&'a str,
+}
+impl MeshIndex<'_>{
+	pub fn file_mesh(content:&str)->MeshIndex<'_>{
+		MeshIndex{
+			mesh_type:MeshType::FileMesh,
+			content,
+		}
+	}
+	pub fn union<'a>(
+		content:&'a str,
+		mesh_data:&'a [u8],
+		physics_data:&'a [u8],
+		size:&rbx_dom_weak::types::Vector3,
+		part_texture_description:RobloxPartDescription,
+	)->MeshIndex<'a>{
+		MeshIndex{
+			mesh_type:MeshType::Union{
+				mesh_data,
+				physics_data,
+				size_float_bits:[size.x.to_bits(),size.y.to_bits(),size.z.to_bits()],
+				part_texture_description,
+			},
+			content,
+		}
+	}
+}
+
+#[derive(Clone)]
+pub struct MeshWithSize{
+	mesh:model::Mesh,
+	size:strafesnet_common::integer::Planar64Vec3,
+}
+impl MeshWithSize{
+	pub(crate) const fn new(
+		mesh:model::Mesh,
+		size:strafesnet_common::integer::Planar64Vec3,
+	)->Self{
+		Self{mesh,size}
+	}
+	pub const fn mesh(&self)->&model::Mesh{
+		&self.mesh
+	}
+	pub const fn size(&self)->strafesnet_common::integer::Planar64Vec3{
+		self.size
+	}
+}
 
 #[derive(Debug)]
 pub enum Error{
-	Planar64Vec3(strafesnet_common::integer::Planar64TryFromFloatError),
-	RbxMesh(rbx_mesh::mesh::Error)
+	NoPolygons,
+	RbxMesh(rbx_mesh::mesh::Error),
 }
 impl std::fmt::Display for Error{
 	fn fmt(&self,f:&mut std::fmt::Formatter<'_>)->std::fmt::Result{
@@ -15,196 +80,143 @@ impl std::fmt::Display for Error{
 }
 impl std::error::Error for Error{}
 
-fn ingest_vertices2<
-	AcquirePosId,
-	AcquireTexId,
-	AcquireNormalId,
-	AcquireColorId,
-	AcquireVertexId,
->(
+fn ingest_vertices2(
 	vertices:Vec<Vertex2>,
-	acquire_pos_id:&mut AcquirePosId,
-	acquire_tex_id:&mut AcquireTexId,
-	acquire_normal_id:&mut AcquireNormalId,
-	acquire_color_id:&mut AcquireColorId,
-	acquire_vertex_id:&mut AcquireVertexId,
-)->Result<HashMap<rbx_mesh::mesh::VertexId2,VertexId>,Error>
-where
-	AcquirePosId:FnMut([f32;3])->Result<PositionId,Error>,
-	AcquireTexId:FnMut([f32;2])->TextureCoordinateId,
-	AcquireNormalId:FnMut([f32;3])->Result<NormalId,Error>,
-	AcquireColorId:FnMut([f32;4])->ColorId,
-	AcquireVertexId:FnMut(IndexedVertex)->VertexId,
-{
+	mb:&mut model::MeshBuilder,
+)->HashMap<rbx_mesh::mesh::VertexId2,VertexId>{
 	//this monster is collecting a map of old_vertices_index -> unique_vertices_index
 	//while also doing the inserting unique entries into lists simultaneously
-	Ok(vertices.into_iter().enumerate().map(|(vertex_id,vertex)|Ok((
+	// vertex positions that fail to convert are DROPPED
+	vertices.into_iter().enumerate().filter_map(|(vertex_id,vertex)|Some((
 		rbx_mesh::mesh::VertexId2(vertex_id as u32),
-		acquire_vertex_id(IndexedVertex{
-			pos:acquire_pos_id(vertex.pos)?,
-			tex:acquire_tex_id(vertex.tex),
-			normal:acquire_normal_id(vertex.norm)?,
-			color:acquire_color_id(vertex.color.map(|f|f as f32/255.0f32))
-		}),
-	))).collect::<Result<_,_>>()?)
+		{
+			let vertex=IndexedVertex{
+				pos:mb.acquire_pos_id(vec3::try_from_f32_array(vertex.pos).ok()?),
+				tex:mb.acquire_tex_id(glam::Vec2::from_array(vertex.tex)),
+				normal:mb.acquire_normal_id(vec3::try_from_f32_array(vertex.norm).ok()?),
+				color:mb.acquire_color_id(glam::Vec4::from_array(vertex.color.map(|f|f as f32/255.0f32))),
+			};
+			mb.acquire_vertex_id(vertex)
+		}
+	))).collect()
 }
-fn ingest_vertices_truncated2<
-	AcquirePosId,
-	AcquireTexId,
-	AcquireNormalId,
-	AcquireVertexId,
->(
+fn ingest_vertices_truncated2(
 	vertices:Vec<Vertex2Truncated>,
-	acquire_pos_id:&mut AcquirePosId,
-	acquire_tex_id:&mut AcquireTexId,
-	acquire_normal_id:&mut AcquireNormalId,
+	mb:&mut model::MeshBuilder,
 	static_color_id:ColorId,//pick one color and fill everything with it
-	acquire_vertex_id:&mut AcquireVertexId,
-)->Result<HashMap<rbx_mesh::mesh::VertexId2,VertexId>,Error>
-where
-	AcquirePosId:FnMut([f32;3])->Result<PositionId,Error>,
-	AcquireTexId:FnMut([f32;2])->TextureCoordinateId,
-	AcquireNormalId:FnMut([f32;3])->Result<NormalId,Error>,
-	AcquireVertexId:FnMut(IndexedVertex)->VertexId,
-{
+)->HashMap<rbx_mesh::mesh::VertexId2,VertexId>{
 	//this monster is collecting a map of old_vertices_index -> unique_vertices_index
 	//while also doing the inserting unique entries into lists simultaneously
-	Ok(vertices.into_iter().enumerate().map(|(vertex_id,vertex)|Ok((
+	// vertex positions that fail to convert are DROPPED
+	vertices.into_iter().enumerate().filter_map(|(vertex_id,vertex)|Some((
 		rbx_mesh::mesh::VertexId2(vertex_id as u32),
-		acquire_vertex_id(IndexedVertex{
-			pos:acquire_pos_id(vertex.pos)?,
-			tex:acquire_tex_id(vertex.tex),
-			normal:acquire_normal_id(vertex.norm)?,
-			color:static_color_id
-		}),
-	))).collect::<Result<_,_>>()?)
+		{
+			let vertex=IndexedVertex{
+				pos:mb.acquire_pos_id(vec3::try_from_f32_array(vertex.pos).ok()?),
+				tex:mb.acquire_tex_id(glam::Vec2::from_array(vertex.tex)),
+				normal:mb.acquire_normal_id(vec3::try_from_f32_array(vertex.norm).ok()?),
+				color:static_color_id,
+			};
+			mb.acquire_vertex_id(vertex)
+		}
+	))).collect()
+}
+
+fn new_polygon_list_checked(list:Vec<model::IndexedVertexList>)->Option<PolygonList>{
+	(!list.is_empty()).then_some(PolygonList::new(list))
 }
 
 fn ingest_faces2_lods3(
 	polygon_groups:&mut Vec<PolygonGroup>,
 	vertex_id_map:&HashMap<rbx_mesh::mesh::VertexId2,VertexId>,
-	faces:&Vec<rbx_mesh::mesh::Face2>,
-	lods:&Vec<rbx_mesh::mesh::Lod3>
+	faces:&[rbx_mesh::mesh::Face2],
+	lods:&[rbx_mesh::mesh::Lod3],
 ){
 	//faces have to be split into polygon groups based on lod
-	polygon_groups.extend(lods.windows(2).map(|lod_pair|
-		PolygonGroup::PolygonList(PolygonList::new(faces[lod_pair[0].0 as usize..lod_pair[1].0 as usize].iter().map(|face|
-			vec![vertex_id_map[&face.0],vertex_id_map[&face.1],vertex_id_map[&face.2]]
-		).collect()))
+	polygon_groups.extend(lods.windows(2).filter_map(|lod_pair|
+		Some(PolygonGroup::PolygonList(new_polygon_list_checked(faces[lod_pair[0].0 as usize..lod_pair[1].0 as usize].iter().filter_map(|rbx_mesh::mesh::Face2(v0,v1,v2)|
+			Some(vec![*vertex_id_map.get(&v0)?,*vertex_id_map.get(&v1)?,*vertex_id_map.get(&v2)?])
+		).collect())?))
 	))
 }
 
-pub fn convert(roblox_mesh_bytes:crate::data::RobloxMeshBytes)->Result<model::Mesh,Error>{
+pub fn convert(roblox_mesh_bytes:&[u8])->Result<MeshWithSize,Error>{
 	//generate that mesh boi
-	let mut unique_pos=Vec::new();
-	let mut pos_id_from=HashMap::new();
-	let mut unique_tex=Vec::new();
-	let mut tex_id_from=HashMap::new();
-	let mut unique_normal=Vec::new();
-	let mut normal_id_from=HashMap::new();
-	let mut unique_color=Vec::new();
-	let mut color_id_from=HashMap::new();
-	let mut unique_vertices=Vec::new();
-	let mut vertex_id_from=HashMap::new();
 	let mut polygon_groups=Vec::new();
-	let mut acquire_pos_id=|pos|{
-		let p=vec3::try_from_f32_array(pos).map_err(Error::Planar64Vec3)?;
-		Ok(PositionId::new(*pos_id_from.entry(p).or_insert_with(||{
-			let pos_id=unique_pos.len();
-			unique_pos.push(p);
-			pos_id
-		}) as u32))
-	};
-	let mut acquire_tex_id=|tex|{
-		let h=bytemuck::cast::<[f32;2],[u32;2]>(tex);
-		TextureCoordinateId::new(*tex_id_from.entry(h).or_insert_with(||{
-			let tex_id=unique_tex.len();
-			unique_tex.push(glam::Vec2::from_array(tex));
-			tex_id
-		}) as u32)
-	};
-	let mut acquire_normal_id=|normal|{
-		let n=vec3::try_from_f32_array(normal).map_err(Error::Planar64Vec3)?;
-		Ok(NormalId::new(*normal_id_from.entry(n).or_insert_with(||{
-			let normal_id=unique_normal.len();
-			unique_normal.push(n);
-			normal_id
-		}) as u32))
-	};
-	let mut acquire_color_id=|color|{
-		let h=bytemuck::cast::<[f32;4],[u32;4]>(color);
-		ColorId::new(*color_id_from.entry(h).or_insert_with(||{
-			let color_id=unique_color.len();
-			unique_color.push(glam::Vec4::from_array(color));
-			color_id
-		}) as u32)
-	};
-	let mut acquire_vertex_id=|vertex:IndexedVertex|{
-		VertexId::new(*vertex_id_from.entry(vertex.clone()).or_insert_with(||{
-			let vertex_id=unique_vertices.len();
-			unique_vertices.push(vertex);
-			vertex_id
-		}) as u32)
-	};
-	match rbx_mesh::read_versioned(roblox_mesh_bytes.cursor()).map_err(Error::RbxMesh)?{
-		rbx_mesh::mesh::VersionedMesh::Version1(mesh)=>{
-			let color_id=acquire_color_id([1.0f32;4]);
-			polygon_groups.push(PolygonGroup::PolygonList(PolygonList::new(mesh.vertices.chunks_exact(3).map(|trip|{
-				let mut ingest_vertex1=|vertex:&rbx_mesh::mesh::Vertex1|Ok(acquire_vertex_id(IndexedVertex{
-					pos:acquire_pos_id(vertex.pos)?,
-					tex:acquire_tex_id([vertex.tex[0],vertex.tex[1]]),
-					normal:acquire_normal_id(vertex.norm)?,
-					color:color_id,
-				}));
-				Ok(vec![ingest_vertex1(&trip[0])?,ingest_vertex1(&trip[1])?,ingest_vertex1(&trip[2])?])
-			}).collect::<Result<_,_>>()?)));
+	let mut mb=model::MeshBuilder::new();
+	match rbx_mesh::read_versioned(std::io::Cursor::new(roblox_mesh_bytes)).map_err(Error::RbxMesh)?{
+		rbx_mesh::mesh::Mesh::V1(mesh)=>{
+			let color_id=mb.acquire_color_id(glam::Vec4::ONE);
+			let polygon_list=new_polygon_list_checked(mesh.vertices.chunks_exact(3).filter_map(|trip|{
+				let mut ingest_vertex1=|vertex:&rbx_mesh::mesh::Vertex1|{
+					let vertex=IndexedVertex{
+						pos:mb.acquire_pos_id(vec3::try_from_f32_array(vertex.pos).ok()?),
+						tex:mb.acquire_tex_id(glam::vec2(vertex.tex[0],vertex.tex[1])),
+						normal:mb.acquire_normal_id(vec3::try_from_f32_array(vertex.norm).ok()?),
+						color:color_id,
+					};
+					Some(mb.acquire_vertex_id(vertex))
+				};
+				Some(vec![ingest_vertex1(&trip[0])?,ingest_vertex1(&trip[1])?,ingest_vertex1(&trip[2])?])
+			}).collect());
+			if let Some(polygon_list)=polygon_list{
+				polygon_groups.push(PolygonGroup::PolygonList(polygon_list));
+			}
 		},
-		rbx_mesh::mesh::VersionedMesh::Version2(mesh)=>{
+		rbx_mesh::mesh::Mesh::V2(mesh)=>{
 			let vertex_id_map=match mesh.header.sizeof_vertex{
 				rbx_mesh::mesh::SizeOfVertex2::Truncated=>{
 					//pick white and make all the vertices white
-					let color_id=acquire_color_id([1.0f32;4]);
-					ingest_vertices_truncated2(mesh.vertices_truncated,&mut acquire_pos_id,&mut acquire_tex_id,&mut acquire_normal_id,color_id,&mut acquire_vertex_id)
+					let color_id=mb.acquire_color_id(glam::Vec4::ONE);
+					ingest_vertices_truncated2(mesh.vertices_truncated,&mut mb,color_id)
 				},
-				rbx_mesh::mesh::SizeOfVertex2::Full=>ingest_vertices2(mesh.vertices,&mut acquire_pos_id,&mut acquire_tex_id,&mut acquire_normal_id,&mut acquire_color_id,&mut acquire_vertex_id),
-			}?;
+				rbx_mesh::mesh::SizeOfVertex2::Full=>ingest_vertices2(mesh.vertices,&mut mb),
+			};
 			//one big happy group for all the faces
-			polygon_groups.push(PolygonGroup::PolygonList(PolygonList::new(mesh.faces.into_iter().map(|face|
-				vec![vertex_id_map[&face.0],vertex_id_map[&face.1],vertex_id_map[&face.2]]
-			).collect())));
+			let polygon_list=new_polygon_list_checked(mesh.faces.into_iter().filter_map(|face|
+				Some(vec![*vertex_id_map.get(&face.0)?,*vertex_id_map.get(&face.1)?,*vertex_id_map.get(&face.2)?])
+			).collect());
+			if let Some(polygon_list)=polygon_list{
+				polygon_groups.push(PolygonGroup::PolygonList(polygon_list));
+			}
 		},
-		rbx_mesh::mesh::VersionedMesh::Version3(mesh)=>{
+		rbx_mesh::mesh::Mesh::V3(mesh)=>{
 			let vertex_id_map=match mesh.header.sizeof_vertex{
 				rbx_mesh::mesh::SizeOfVertex2::Truncated=>{
-					let color_id=acquire_color_id([1.0f32;4]);
-					ingest_vertices_truncated2(mesh.vertices_truncated,&mut acquire_pos_id,&mut acquire_tex_id,&mut acquire_normal_id,color_id,&mut acquire_vertex_id)
+					let color_id=mb.acquire_color_id(glam::Vec4::ONE);
+					ingest_vertices_truncated2(mesh.vertices_truncated,&mut mb,color_id)
 				},
-				rbx_mesh::mesh::SizeOfVertex2::Full=>ingest_vertices2(mesh.vertices,&mut acquire_pos_id,&mut acquire_tex_id,&mut acquire_normal_id,&mut acquire_color_id,&mut acquire_vertex_id),
-			}?;
+				rbx_mesh::mesh::SizeOfVertex2::Full=>ingest_vertices2(mesh.vertices,&mut mb),
+			};
 			ingest_faces2_lods3(&mut polygon_groups,&vertex_id_map,&mesh.faces,&mesh.lods);
 		},
-		rbx_mesh::mesh::VersionedMesh::Version4(mesh)=>{
-			let vertex_id_map=ingest_vertices2(
-				mesh.vertices,&mut acquire_pos_id,&mut acquire_tex_id,&mut acquire_normal_id,&mut acquire_color_id,&mut acquire_vertex_id
-			)?;
+		rbx_mesh::mesh::Mesh::V4(mesh)=>{
+			let vertex_id_map=ingest_vertices2(mesh.vertices,&mut mb);
 			ingest_faces2_lods3(&mut polygon_groups,&vertex_id_map,&mesh.faces,&mesh.lods);
 		},
-		rbx_mesh::mesh::VersionedMesh::Version5(mesh)=>{
-			let vertex_id_map=ingest_vertices2(
-				mesh.vertices,&mut acquire_pos_id,&mut acquire_tex_id,&mut acquire_normal_id,&mut acquire_color_id,&mut acquire_vertex_id
-			)?;
+		rbx_mesh::mesh::Mesh::V5(mesh)=>{
+			let vertex_id_map=ingest_vertices2(mesh.vertices,&mut mb);
 			ingest_faces2_lods3(&mut polygon_groups,&vertex_id_map,&mesh.faces,&mesh.lods);
 		},
 	}
-	Ok(model::Mesh{
-		unique_pos,
-		unique_normal,
-		unique_tex,
-		unique_color,
-		unique_vertices,
+	if polygon_groups.is_empty(){
+		return Err(Error::NoPolygons);
+	}
+	let mesh=mb.build(
 		polygon_groups,
 		//these should probably be moved to the model...
-		graphics_groups:Vec::new(),
-		physics_groups:Vec::new(),
-	})
+		//but what if models want to use the same texture
+		vec![model::IndexedGraphicsGroup{
+			render:RenderConfigId::new(0),
+			//the lowest lod is highest quality
+			groups:vec![model::PolygonGroupId::new(0)]
+		}],
+		//disable physics
+		Vec::new(),
+	);
+	let mut aabb=Aabb::default();
+	for &point in &mesh.unique_pos{
+		aabb.grow(point);
+	}
+	Ok(MeshWithSize{mesh,size:aabb.size()})
 }

lib/rbx_loader/src/primitives.rs

@@ -1,5 +1,6 @@
-use strafesnet_common::model::{Color4,TextureCoordinate,Mesh,IndexedGraphicsGroup,IndexedPhysicsGroup,IndexedVertex,PolygonGroupId,PolygonGroup,PolygonList,IndexedVertexList,PositionId,TextureCoordinateId,NormalId,ColorId,VertexId,RenderConfigId};
-use strafesnet_common::integer::{vec3,Planar64Vec3};
+use crate::rbx::{RobloxPartDescription,RobloxWedgeDescription,RobloxCornerWedgeDescription};
+use strafesnet_common::model::{Color4,TextureCoordinate,Mesh,MeshBuilder,IndexedGraphicsGroup,IndexedPhysicsGroup,IndexedVertex,PolygonGroupId,PolygonGroup,PolygonList,PositionId,TextureCoordinateId,NormalId,ColorId,VertexId,RenderConfigId};
+use strafesnet_common::integer::{vec3,Planar64,Planar64Vec3};
 
 #[derive(Debug)]
 pub enum Primitives{
@@ -9,7 +10,22 @@ pub enum Primitives{
 	Wedge,
 	CornerWedge,
 }
-#[derive(Hash,PartialEq,Eq)]
+#[derive(Debug)]
+pub struct PrimitivesError;
+impl TryFrom<u32> for Primitives{
+	type Error=PrimitivesError;
+	fn try_from(value:u32)->Result<Self,Self::Error>{
+		match value{
+			0=>Ok(Primitives::Sphere),
+			1=>Ok(Primitives::Cube),
+			2=>Ok(Primitives::Cylinder),
+			3=>Ok(Primitives::Wedge),
+			4=>Ok(Primitives::CornerWedge),
+			_=>Err(PrimitivesError),
+		}
+	}
+}
+#[derive(Clone,Copy,Hash,PartialEq,Eq)]
 pub enum CubeFace{
 	Right,
 	Top,
@@ -18,13 +34,29 @@ pub enum CubeFace{
 	Bottom,
 	Front,
 }
+#[derive(Debug)]
+pub struct CubeFaceError;
+impl TryFrom<u32> for CubeFace{
+	type Error=CubeFaceError;
+	fn try_from(value:u32)->Result<Self,Self::Error>{
+		match value{
+			0=>Ok(CubeFace::Right),
+			1=>Ok(CubeFace::Top),
+			2=>Ok(CubeFace::Back),
+			3=>Ok(CubeFace::Left),
+			4=>Ok(CubeFace::Bottom),
+			5=>Ok(CubeFace::Front),
+			_=>Err(CubeFaceError),
+		}
+	}
+}
 const CUBE_DEFAULT_TEXTURE_COORDS:[TextureCoordinate;4]=[
 	TextureCoordinate::new(0.0,0.0),
 	TextureCoordinate::new(1.0,0.0),
 	TextureCoordinate::new(1.0,1.0),
 	TextureCoordinate::new(0.0,1.0),
 ];
-const CUBE_DEFAULT_VERTICES:[Planar64Vec3;8]=[
+pub const CUBE_DEFAULT_VERTICES:[Planar64Vec3;8]=[
 	vec3::int(-1,-1, 1),//0 left bottom back
 	vec3::int( 1,-1, 1),//1 right bottom back
 	vec3::int( 1, 1, 1),//2 right top back
@@ -34,7 +66,7 @@ const CUBE_DEFAULT_VERTICES:[Planar64Vec3;8]=[
 	vec3::int( 1,-1,-1),//6 right bottom front
 	vec3::int(-1,-1,-1),//7 left bottom front
 ];
-const CUBE_DEFAULT_NORMALS:[Planar64Vec3;6]=[
+pub const CUBE_DEFAULT_NORMALS:[Planar64Vec3;6]=[
 	vec3::int( 1, 0, 0),//CubeFace::Right
 	vec3::int( 0, 1, 0),//CubeFace::Top
 	vec3::int( 0, 0, 1),//CubeFace::Back
@@ -42,154 +74,36 @@ const CUBE_DEFAULT_NORMALS:[Planar64Vec3;6]=[
 	vec3::int( 0,-1, 0),//CubeFace::Bottom
 	vec3::int( 0, 0,-1),//CubeFace::Front
 ];
-const CUBE_DEFAULT_POLYS:[[[u32;3];4];6]=[
-	// right (1, 0, 0)
-	[
-		[6,2,0],//[vertex,tex,norm]
-		[5,1,0],
-		[2,0,0],
-		[1,3,0],
-	],
-	// top (0, 1, 0)
-	[
-		[5,3,1],
-		[4,2,1],
-		[3,1,1],
-		[2,0,1],
-	],
-	// back (0, 0, 1)
-	[
-		[0,3,2],
-		[1,2,2],
-		[2,1,2],
-		[3,0,2],
-	],
-	// left (-1, 0, 0)
-	[
-		[0,2,3],
-		[3,1,3],
-		[4,0,3],
-		[7,3,3],
-	],
-	// bottom (0,-1, 0)
-	[
-		[1,1,4],
-		[0,0,4],
-		[7,3,4],
-		[6,2,4],
-	],
-	// front (0, 0,-1)
-	[
-		[4,1,5],
-		[5,0,5],
-		[6,3,5],
-		[7,2,5],
-	],
-];
 
-#[derive(Hash,PartialEq,Eq)]
-pub enum WedgeFace{
-	Right,
-	TopFront,
-	Back,
-	Left,
-	Bottom,
-}
-const WEDGE_DEFAULT_NORMALS:[Planar64Vec3;5]=[
-	vec3::int( 1, 0, 0),//Wedge::Right
-	vec3::int( 0, 1,-1),//Wedge::TopFront
-	vec3::int( 0, 0, 1),//Wedge::Back
-	vec3::int(-1, 0, 0),//Wedge::Left
-	vec3::int( 0,-1, 0),//Wedge::Bottom
-];
-/*
-local cornerWedgeVerticies = {
-	Vector3.new(-1/2,-1/2,-1/2),7
-	Vector3.new(-1/2,-1/2, 1/2),0
-	Vector3.new( 1/2,-1/2,-1/2),6
-	Vector3.new( 1/2,-1/2, 1/2),1
-	Vector3.new( 1/2, 1/2,-1/2),5
-}
-*/
-#[derive(Hash,PartialEq,Eq)]
-pub enum CornerWedgeFace{
-	Right,
-	TopBack,
-	TopLeft,
-	Bottom,
-	Front,
-}
-const CORNERWEDGE_DEFAULT_NORMALS:[Planar64Vec3;5]=[
-	vec3::int( 1, 0, 0),//CornerWedge::Right
-	vec3::int( 0, 1, 1),//CornerWedge::BackTop
-	vec3::int(-1, 1, 0),//CornerWedge::LeftTop
-	vec3::int( 0,-1, 0),//CornerWedge::Bottom
-	vec3::int( 0, 0,-1),//CornerWedge::Front
-];
-pub fn unit_sphere(render:RenderConfigId)->Mesh{
-	unit_cube(render)
-}
-#[derive(Default)]
-pub struct CubeFaceDescription([Option<FaceDescription>;6]);
+pub struct CubeFaceDescription([FaceDescription;Self::FACES]);
 impl CubeFaceDescription{
-	pub fn insert(&mut self,index:CubeFace,value:FaceDescription){
-		self.0[index as usize]=Some(value);
-	}
-	pub fn pairs(self)->std::iter::FilterMap<std::iter::Enumerate<std::array::IntoIter<Option<FaceDescription>,6>>,impl FnMut((usize,Option<FaceDescription>))->Option<(usize,FaceDescription)>>{
-		self.0.into_iter().enumerate().filter_map(|v|v.1.map(|u|(v.0,u)))
+	pub const FACES:usize=6;
+	pub fn new(RobloxPartDescription(part_description):RobloxPartDescription,textureless_render_id:RenderConfigId)->Self{
+		Self(part_description.map(|face_description|match face_description{
+			Some(roblox_texture_transform)=>roblox_texture_transform.to_face_description(),
+			None=>FaceDescription::new_with_render_id(textureless_render_id),
+		}))
 	}
 }
-pub fn unit_cube(render:RenderConfigId)->Mesh{
-	let mut t=CubeFaceDescription::default();
-	t.insert(CubeFace::Right,FaceDescription::new_with_render_id(render));
-	t.insert(CubeFace::Top,FaceDescription::new_with_render_id(render));
-	t.insert(CubeFace::Back,FaceDescription::new_with_render_id(render));
-	t.insert(CubeFace::Left,FaceDescription::new_with_render_id(render));
-	t.insert(CubeFace::Bottom,FaceDescription::new_with_render_id(render));
-	t.insert(CubeFace::Front,FaceDescription::new_with_render_id(render));
-	generate_partial_unit_cube(t)
-}
-pub fn unit_cylinder(render:RenderConfigId)->Mesh{
-	//lmao
-	unit_cube(render)
-}
-#[derive(Default)]
-pub struct WedgeFaceDescription([Option<FaceDescription>;5]);
+pub struct WedgeFaceDescription([FaceDescription;Self::FACES]);
 impl WedgeFaceDescription{
-	pub fn insert(&mut self,index:WedgeFace,value:FaceDescription){
-		self.0[index as usize]=Some(value);
-	}
-	pub fn pairs(self)->std::iter::FilterMap<std::iter::Enumerate<std::array::IntoIter<Option<FaceDescription>,5>>,impl FnMut((usize,Option<FaceDescription>))->Option<(usize,FaceDescription)>>{
-		self.0.into_iter().enumerate().filter_map(|v|v.1.map(|u|(v.0,u)))
+	pub const FACES:usize=5;
+	pub fn new(RobloxWedgeDescription(part_description):RobloxWedgeDescription,textureless_render_id:RenderConfigId)->Self{
+		Self(part_description.map(|face_description|match face_description{
+			Some(roblox_texture_transform)=>roblox_texture_transform.to_face_description(),
+			None=>FaceDescription::new_with_render_id(textureless_render_id),
+		}))
 	}
 }
-pub fn unit_wedge(render:RenderConfigId)->Mesh{
-	let mut t=WedgeFaceDescription::default();
-	t.insert(WedgeFace::Right,FaceDescription::new_with_render_id(render));
-	t.insert(WedgeFace::TopFront,FaceDescription::new_with_render_id(render));
-	t.insert(WedgeFace::Back,FaceDescription::new_with_render_id(render));
-	t.insert(WedgeFace::Left,FaceDescription::new_with_render_id(render));
-	t.insert(WedgeFace::Bottom,FaceDescription::new_with_render_id(render));
-	generate_partial_unit_wedge(t)
-}
-#[derive(Default)]
-pub struct CornerWedgeFaceDescription([Option<FaceDescription>;5]);
+pub struct CornerWedgeFaceDescription([FaceDescription;Self::FACES]);
 impl CornerWedgeFaceDescription{
-	pub fn insert(&mut self,index:CornerWedgeFace,value:FaceDescription){
-		self.0[index as usize]=Some(value);
+	pub const FACES:usize=5;
+	pub fn new(RobloxCornerWedgeDescription(part_description):RobloxCornerWedgeDescription,textureless_render_id:RenderConfigId)->Self{
+		Self(part_description.map(|face_description|match face_description{
+			Some(roblox_texture_transform)=>roblox_texture_transform.to_face_description(),
+			None=>FaceDescription::new_with_render_id(textureless_render_id),
+		}))
 	}
-	pub fn pairs(self)->std::iter::FilterMap<std::iter::Enumerate<std::array::IntoIter<Option<FaceDescription>,5>>,impl FnMut((usize,Option<FaceDescription>))->Option<(usize,FaceDescription)>>{
-		self.0.into_iter().enumerate().filter_map(|v|v.1.map(|u|(v.0,u)))
-	}
-}
-pub fn unit_cornerwedge(render:RenderConfigId)->Mesh{
-	let mut t=CornerWedgeFaceDescription::default();
-	t.insert(CornerWedgeFace::Right,FaceDescription::new_with_render_id(render));
-	t.insert(CornerWedgeFace::TopBack,FaceDescription::new_with_render_id(render));
-	t.insert(CornerWedgeFace::TopLeft,FaceDescription::new_with_render_id(render));
-	t.insert(CornerWedgeFace::Bottom,FaceDescription::new_with_render_id(render));
-	t.insert(CornerWedgeFace::Front,FaceDescription::new_with_render_id(render));
-	generate_partial_unit_cornerwedge(t)
 }
 
 #[derive(Clone)]
@@ -199,7 +113,7 @@ pub struct FaceDescription{
 	pub color:Color4,
 }
 impl FaceDescription{
-	pub fn new_with_render_id(render:RenderConfigId)->Self {
+	pub fn new_with_render_id(render:RenderConfigId)->Self{
 		Self{
 			render,
 			transform:glam::Affine2::IDENTITY,
@@ -207,7 +121,51 @@ impl FaceDescription{
 		}
 	}
 }
-pub fn generate_partial_unit_cube(face_descriptions:CubeFaceDescription)->Mesh{
+pub const CUBE_DEFAULT_POLYS:[[[u32;2];4];6]=[
+		// right (1, 0, 0)
+		[
+			[6,2],//[vertex,tex]
+			[5,1],
+			[2,0],
+			[1,3],
+		],
+		// top (0, 1, 0)
+		[
+			[5,3],
+			[4,2],
+			[3,1],
+			[2,0],
+		],
+		// back (0, 0, 1)
+		[
+			[0,3],
+			[1,2],
+			[2,1],
+			[3,0],
+		],
+		// left (-1, 0, 0)
+		[
+			[0,2],
+			[3,1],
+			[4,0],
+			[7,3],
+		],
+		// bottom (0,-1, 0)
+		[
+			[1,1],
+			[0,0],
+			[7,3],
+			[6,2],
+		],
+		// front (0, 0,-1)
+		[
+			[4,1],
+			[5,0],
+			[6,3],
+			[7,2],
+		],
+	];
+pub fn unit_cube(CubeFaceDescription(face_descriptions):CubeFaceDescription)->Mesh{
 	let mut generated_pos=Vec::new();
 	let mut generated_tex=Vec::new();
 	let mut generated_normal=Vec::new();
@@ -218,7 +176,7 @@ pub fn generate_partial_unit_cube(face_descriptions:CubeFaceDescription)->Mesh{
 	let mut physics_group=IndexedPhysicsGroup::default();
 	let mut transforms=Vec::new();
 	//note that on a cube every vertex is guaranteed to be unique, so there's no need to hash them against existing vertices.
-	for (face_id,face_description) in face_descriptions.pairs(){
+	for (face_id,face_description) in face_descriptions.into_iter().enumerate(){
 		//assume that scanning short lists is faster than hashing.
 		let transform_index=if let Some(transform_index)=transforms.iter().position(|&transform|transform==face_description.transform){
 			transform_index
@@ -245,8 +203,8 @@ pub fn generate_partial_unit_cube(face_descriptions:CubeFaceDescription)->Mesh{
 		//push vertices as they are needed
 		let group_id=PolygonGroupId::new(polygon_groups.len() as u32);
 		polygon_groups.push(PolygonGroup::PolygonList(PolygonList::new(vec![
-			CUBE_DEFAULT_POLYS[face_id].map(|tup|{
-				let pos=CUBE_DEFAULT_VERTICES[tup[0] as usize];
+			CUBE_DEFAULT_POLYS[face_id].map(|[pos_id,tex_id]|{
+				let pos=CUBE_DEFAULT_VERTICES[pos_id as usize];
 				let pos_index=if let Some(pos_index)=generated_pos.iter().position(|&p|p==pos){
 					pos_index
 				}else{
@@ -258,7 +216,7 @@ pub fn generate_partial_unit_cube(face_descriptions:CubeFaceDescription)->Mesh{
 				//always push vertex
 				let vertex=IndexedVertex{
 					pos:PositionId::new(pos_index),
-					tex:TextureCoordinateId::new(tup[1]+4*transform_index),
+					tex:TextureCoordinateId::new(tex_id+4*transform_index),
 					normal:NormalId::new(normal_index),
 					color:ColorId::new(color_index),
 				};
@@ -285,42 +243,49 @@ pub fn generate_partial_unit_cube(face_descriptions:CubeFaceDescription)->Mesh{
 	}
 }
 //don't think too hard about the copy paste because this is all going into the map tool eventually...
-pub fn generate_partial_unit_wedge(face_descriptions:WedgeFaceDescription)->Mesh{
-	let wedge_default_polys=[
+pub fn unit_wedge(WedgeFaceDescription(face_descriptions):WedgeFaceDescription)->Mesh{
+	const WEDGE_DEFAULT_POLYS:[&[[u32;2]];5]=[
 		// right (1, 0, 0)
-		vec![
-			[6,2,0],//[vertex,tex,norm]
-			[2,0,0],
-			[1,3,0],
+		&[
+			[6,2],//[vertex,tex]
+			[2,0],
+			[1,3],
 		],
 		// FrontTop (0, 1, -1)
-		vec![
-			[3,1,1],
-			[2,0,1],
-			[6,3,1],
-			[7,2,1],
+		&[
+			[3,1],
+			[2,0],
+			[6,3],
+			[7,2],
 		],
 		// back (0, 0, 1)
-		vec![
-			[0,3,2],
-			[1,2,2],
-			[2,1,2],
-			[3,0,2],
+		&[
+			[0,3],
+			[1,2],
+			[2,1],
+			[3,0],
 		],
 		// left (-1, 0, 0)
-		vec![
-			[0,2,3],
-			[3,1,3],
-			[7,3,3],
+		&[
+			[0,2],
+			[3,1],
+			[7,3],
 		],
 		// bottom (0,-1, 0)
-		vec![
-			[1,1,4],
-			[0,0,4],
-			[7,3,4],
-			[6,2,4],
+		&[
+			[1,1],
+			[0,0],
+			[7,3],
+			[6,2],
 		],
 	];
+	const WEDGE_DEFAULT_NORMALS:[Planar64Vec3;5]=[
+		vec3::int( 1, 0, 0),//Wedge::Right
+		vec3::int( 0, 1,-1),//Wedge::TopFront
+		vec3::int( 0, 0, 1),//Wedge::Back
+		vec3::int(-1, 0, 0),//Wedge::Left
+		vec3::int( 0,-1, 0),//Wedge::Bottom
+	];
 	let mut generated_pos=Vec::new();
 	let mut generated_tex=Vec::new();
 	let mut generated_normal=Vec::new();
@@ -331,7 +296,7 @@ pub fn generate_partial_unit_wedge(face_descriptions:WedgeFaceDescription)->Mesh
 	let mut physics_group=IndexedPhysicsGroup::default();
 	let mut transforms=Vec::new();
 	//note that on a cube every vertex is guaranteed to be unique, so there's no need to hash them against existing vertices.
-	for (face_id,face_description) in face_descriptions.pairs(){
+	for (face_id,face_description) in face_descriptions.into_iter().enumerate(){
 		//assume that scanning short lists is faster than hashing.
 		let transform_index=if let Some(transform_index)=transforms.iter().position(|&transform|transform==face_description.transform){
 			transform_index
@@ -358,8 +323,8 @@ pub fn generate_partial_unit_wedge(face_descriptions:WedgeFaceDescription)->Mesh
 		//push vertices as they are needed
 		let group_id=PolygonGroupId::new(polygon_groups.len() as u32);
 		polygon_groups.push(PolygonGroup::PolygonList(PolygonList::new(vec![
-			wedge_default_polys[face_id].iter().map(|tup|{
-				let pos=CUBE_DEFAULT_VERTICES[tup[0] as usize];
+			WEDGE_DEFAULT_POLYS[face_id].iter().map(|&[pos_id,tex_id]|{
+				let pos=CUBE_DEFAULT_VERTICES[pos_id as usize];
 				let pos_index=if let Some(pos_index)=generated_pos.iter().position(|&p|p==pos){
 					pos_index
 				}else{
@@ -371,7 +336,7 @@ pub fn generate_partial_unit_wedge(face_descriptions:WedgeFaceDescription)->Mesh
 				//always push vertex
 				let vertex=IndexedVertex{
 					pos:PositionId::new(pos_index),
-					tex:TextureCoordinateId::new(tup[1]+4*transform_index),
+					tex:TextureCoordinateId::new(tex_id+4*transform_index),
 					normal:NormalId::new(normal_index),
 					color:ColorId::new(color_index),
 				};
@@ -398,40 +363,47 @@ pub fn generate_partial_unit_wedge(face_descriptions:WedgeFaceDescription)->Mesh
 	}
 }
 
-pub fn generate_partial_unit_cornerwedge(face_descriptions:CornerWedgeFaceDescription)->Mesh{
-	let cornerwedge_default_polys=[
+pub fn unit_cornerwedge(CornerWedgeFaceDescription(face_descriptions):CornerWedgeFaceDescription)->Mesh{
+	const CORNERWEDGE_DEFAULT_POLYS:[&[[u32;2]];5]=[
 		// right (1, 0, 0)
-		vec![
-			[6,2,0],//[vertex,tex,norm]
-			[5,1,0],
-			[1,3,0],
+		&[
+			[6,2],//[vertex,tex]
+			[5,1],
+			[1,3],
 		],
 		// BackTop (0, 1, 1)
-		vec![
-			[5,3,1],
-			[0,1,1],
-			[1,0,1],
+		&[
+			[5,3],
+			[0,1],
+			[1,0],
 		],
 		// LeftTop (-1, 1, 0)
-		vec![
-			[5,3,2],
-			[7,2,2],
-			[0,1,2],
+		&[
+			[5,3],
+			[7,2],
+			[0,1],
 		],
 		// bottom (0,-1, 0)
-		vec![
-			[1,1,3],
-			[0,0,3],
-			[7,3,3],
-			[6,2,3],
+		&[
+			[1,1],
+			[0,0],
+			[7,3],
+			[6,2],
 		],
 		// front (0, 0,-1)
-		vec![
-			[5,0,4],
-			[6,3,4],
-			[7,2,4],
+		&[
+			[5,0],
+			[6,3],
+			[7,2],
 		],
 	];
+	const CORNERWEDGE_DEFAULT_NORMALS:[Planar64Vec3;5]=[
+		vec3::int( 1, 0, 0),//CornerWedge::Right
+		vec3::int( 0, 1, 1),//CornerWedge::BackTop
+		vec3::int(-1, 1, 0),//CornerWedge::LeftTop
+		vec3::int( 0,-1, 0),//CornerWedge::Bottom
+		vec3::int( 0, 0,-1),//CornerWedge::Front
+	];
 	let mut generated_pos=Vec::new();
 	let mut generated_tex=Vec::new();
 	let mut generated_normal=Vec::new();
@@ -442,7 +414,7 @@ pub fn generate_partial_unit_cornerwedge(face_descriptions:CornerWedgeFaceDescri
 	let mut physics_group=IndexedPhysicsGroup::default();
 	let mut transforms=Vec::new();
 	//note that on a cube every vertex is guaranteed to be unique, so there's no need to hash them against existing vertices.
-	for (face_id,face_description) in face_descriptions.pairs(){
+	for (face_id,face_description) in face_descriptions.into_iter().enumerate(){
 		//assume that scanning short lists is faster than hashing.
 		let transform_index=if let Some(transform_index)=transforms.iter().position(|&transform|transform==face_description.transform){
 			transform_index
@@ -469,8 +441,8 @@ pub fn generate_partial_unit_cornerwedge(face_descriptions:CornerWedgeFaceDescri
 		//push vertices as they are needed
 		let group_id=PolygonGroupId::new(polygon_groups.len() as u32);
 		polygon_groups.push(PolygonGroup::PolygonList(PolygonList::new(vec![
-			cornerwedge_default_polys[face_id].iter().map(|tup|{
-				let pos=CUBE_DEFAULT_VERTICES[tup[0] as usize];
+			CORNERWEDGE_DEFAULT_POLYS[face_id].iter().map(|&[pos_id,tex_id]|{
+				let pos=CUBE_DEFAULT_VERTICES[pos_id as usize];
 				let pos_index=if let Some(pos_index)=generated_pos.iter().position(|&p|p==pos){
 					pos_index
 				}else{
@@ -482,7 +454,7 @@ pub fn generate_partial_unit_cornerwedge(face_descriptions:CornerWedgeFaceDescri
 				//always push vertex
 				let vertex=IndexedVertex{
 					pos:PositionId::new(pos_index),
-					tex:TextureCoordinateId::new(tup[1]+4*transform_index),
+					tex:TextureCoordinateId::new(tex_id+4*transform_index),
 					normal:NormalId::new(normal_index),
 					color:ColorId::new(color_index),
 				};
@@ -508,3 +480,133 @@ pub fn generate_partial_unit_cornerwedge(face_descriptions:CornerWedgeFaceDescri
 		physics_groups:vec![physics_group],
 	}
 }
+
+// TODO: fix face texture orientation
+pub fn unit_cylinder(face_descriptions:CubeFaceDescription)->Mesh{
+	// cylinder is oriented about the x axis
+	// roblox cylinders use projected grid coordinates
+	/// how many grid coordinates to use (positive and negative)
+	const GON:i32=3;
+	/// grid perimeter
+	const POINTS:[[i32;2];4*2*GON as usize]=const{
+		let mut points=[[0;2];{4*2*GON as usize}];
+		let mut i=-GON;
+		while i<GON{
+			points[(i+GON) as usize]=[i,GON];
+			points[(i+GON+1*2*GON) as usize]=[GON,-i];
+			points[(i+GON+2*2*GON) as usize]=[-i,-GON];
+			points[(i+GON+3*2*GON) as usize]=[-GON,i];
+			i+=1;
+		}
+		points
+	};
+
+	let mut mb=MeshBuilder::new();
+	let mut polygon_groups=Vec::with_capacity(CubeFaceDescription::FACES);
+	let mut graphics_groups=Vec::with_capacity(CubeFaceDescription::FACES);
+	let mut physics_group=IndexedPhysicsGroup{groups:Vec::with_capacity(CubeFaceDescription::FACES)};
+	let CubeFaceDescription([right,top,back,left,bottom,front])=face_descriptions;
+
+	macro_rules! end_face{
+		($face_description:expr,$end:expr,$iter:expr)=>{
+			let normal=mb.acquire_normal_id($end);
+			let color=mb.acquire_color_id($face_description.color);
+
+			// single polygon for physics
+			let polygon:Vec<_>=$iter.map(|[x,y]|{
+				let tex=mb.acquire_tex_id(
+					$face_description.transform.transform_point2(
+						(glam::vec2(-x as f32,y as f32).normalize()+1.0)/2.0
+					)
+				);
+				let pos=mb.acquire_pos_id($end+vec3::int(0,-x,y).with_length(Planar64::ONE).divide().wrap_64());
+				mb.acquire_vertex_id(IndexedVertex{pos,tex,normal,color})
+			}).collect();
+
+			// fanned polygons for graphics
+			let pos=mb.acquire_pos_id($end);
+			let tex=mb.acquire_tex_id($face_description.transform.transform_point2(glam::Vec2::ONE/2.0));
+			let center=mb.acquire_vertex_id(IndexedVertex{pos,tex,normal,color});
+			let polygon_list=(0..POINTS.len()).map(|i|
+				vec![center,polygon[i],polygon[(i+1)%POINTS.len()]]
+			).collect();
+
+			// end face graphics
+			let group_id=PolygonGroupId::new(polygon_groups.len() as u32);
+			polygon_groups.push(PolygonGroup::PolygonList(PolygonList::new(polygon_list)));
+			graphics_groups.push(IndexedGraphicsGroup{
+				render:$face_description.render,
+				groups:vec![group_id],
+			});
+
+			// end face physics
+			let polygon_list=vec![polygon];
+
+			let group_id=PolygonGroupId::new(polygon_groups.len() as u32);
+			polygon_groups.push(PolygonGroup::PolygonList(PolygonList::new(polygon_list)));
+			physics_group.groups.push(group_id);
+		}
+	}
+	macro_rules! tex{
+		($face_description:expr,$tex:expr)=>{{
+			let [x,y]=$tex;
+			$face_description.transform.transform_point2(
+				glam::vec2((x+GON) as f32,(y+GON) as f32)/(2*GON) as f32
+			)
+		}};
+	}
+	macro_rules! barrel_face{
+		($face_description:expr,$loop:ident,$lo_dir:expr,$hi_dir:expr,$tex_0:expr,$tex_1:expr,$tex_2:expr,$tex_3:expr)=>{
+			let mut polygon_list=Vec::with_capacity(CubeFaceDescription::FACES);
+			for $loop in -GON..GON{
+				// lo Z
+				let lz_dir=$lo_dir.with_length(Planar64::ONE).divide().wrap_64();
+				// hi Z
+				let hz_dir=$hi_dir.with_length(Planar64::ONE).divide().wrap_64();
+
+				// pos
+				let lx_lz_pos=mb.acquire_pos_id(vec3::NEG_X+lz_dir);
+				let lx_hz_pos=mb.acquire_pos_id(vec3::NEG_X+hz_dir);
+				let hx_hz_pos=mb.acquire_pos_id(vec3::X+hz_dir);
+				let hx_lz_pos=mb.acquire_pos_id(vec3::X+lz_dir);
+				// tex
+				let lx_lz_tex=mb.acquire_tex_id(tex!($face_description,$tex_0));
+				let lx_hz_tex=mb.acquire_tex_id(tex!($face_description,$tex_1));
+				let hx_hz_tex=mb.acquire_tex_id(tex!($face_description,$tex_2));
+				let hx_lz_tex=mb.acquire_tex_id(tex!($face_description,$tex_3));
+				// norm
+				let lz_norm=mb.acquire_normal_id(lz_dir);
+				let hz_norm=mb.acquire_normal_id(hz_dir);
+				// color
+				let color=mb.acquire_color_id($face_description.color);
+
+				polygon_list.push(vec![
+					mb.acquire_vertex_id(IndexedVertex{pos:lx_lz_pos,tex:lx_lz_tex,normal:lz_norm,color}),
+					mb.acquire_vertex_id(IndexedVertex{pos:lx_hz_pos,tex:lx_hz_tex,normal:hz_norm,color}),
+					mb.acquire_vertex_id(IndexedVertex{pos:hx_hz_pos,tex:hx_hz_tex,normal:hz_norm,color}),
+					mb.acquire_vertex_id(IndexedVertex{pos:hx_lz_pos,tex:hx_lz_tex,normal:lz_norm,color}),
+				]);
+			}
+
+			// push face
+			let group_id=PolygonGroupId::new(polygon_groups.len() as u32);
+			polygon_groups.push(PolygonGroup::PolygonList(PolygonList::new(polygon_list)));
+			graphics_groups.push(IndexedGraphicsGroup{
+				render:$face_description.render,
+				groups:vec![group_id],
+			});
+			physics_group.groups.push(group_id);
+		};
+	}
+
+	end_face!(right, vec3::X,POINTS.into_iter());
+	barrel_face!(top, z,vec3::int(0,GON,z),vec3::int(0,GON,z+1), [GON,z],[GON,z+1],[-GON,z+1],[-GON,z]);
+	barrel_face!(back, y,vec3::int(0,y+1,GON),vec3::int(0,y,GON), [GON,y+1],[GON,y],[-GON,y],[-GON,y+1]);
+	end_face!(left, vec3::NEG_X,POINTS.into_iter().rev());
+	barrel_face!(bottom, z,vec3::int(0,-GON,z+1),vec3::int(0,-GON,z), [-GON,z+1],[-GON,z],[GON,z],[GON,z+1]);
+	barrel_face!(front, y,vec3::int(0,y,-GON),vec3::int(0,y+1,-GON), [-GON,y],[-GON,y+1],[GON,y+1],[GON,y]);
+
+	let physics_groups=vec![physics_group];
+
+	mb.build(polygon_groups,graphics_groups,physics_groups)
+}