Crate wde_wgpu 

WaterDropEngine’s wde-wgpu crate is a lightweight layer over wgpu that keeps the low-level power of WebGPU while offering opinionated building blocks for buffers, textures, render/compute pipelines, and pass orchestration. All submodules are re-exported at the crate root (e.g. wde_wgpu::buffer, wde_wgpu::command_buffer) to keep imports short and examples consistent.

Architecture

• Instance and surface: instance::create_instance produces a instance::RenderInstance holding the wgpu::Device, Queue, and optional Surface. instance::setup_surface stores a [wgpu::SurfaceConfiguration] once the window size is known and instance::resize reapplies it on swapchain loss or window resize.
• Resources: buffer::Buffer and texture::Texture wrap GPU memory and add convenience copy helpers plus defaults for common usages (render targets, depth).
• Pipelines: render_pipeline::RenderPipeline and compute_pipeline::ComputePipeline compile WGSL source, build pipeline layouts (bind groups + push constants), and expose is_initialized + getters for the underlying wgpu objects.
• Pass orchestration: command_buffer::CommandBuffer records GPU work. It spawns render_pass::RenderPassInstance and compute_pass::WComputePass with guard rails that check for missing pipelines/buffers before issuing draws or dispatches.
• Bind groups: bind_group helps assemble layouts and bind groups that match WGSL declarations.
• Vertex helpers: vertex::Vertex provides a canonical position/uv/normal layout and a ready-to-use wgpu::VertexBufferLayout.

Core usage patterns

• Configure a surface once, then react to instance::RenderEvent::Resize with instance::resize.
• Build resources up-front (Buffer, Texture), then stage per-frame uploads via buffer::Buffer::write or queue.write_texture wrappers.
• Keep pipeline creation deterministic: set shaders, bind group layouts, push constants, then call init and gate draws/dispatches on is_initialized.
• Record work inside a CommandBuffer; each render/compute pass enforces that required state (pipeline, vertex/index buffers) is set before issuing commands.

Modules

• instance: device, queue, surface, events, presentation helpers.
• buffer / texture: GPU buffers and 2D textures with helper views/samplers.
• render_pipeline / compute_pipeline: WGSL compilation + pipeline layout creation.
• render_pass / compute_pass / command_buffer: pass-level helpers with safety checks.
• bind_group: bind group layout builders and bind group creation.
• vertex: canonical vertex layout used across examples.

Examples and further reading

• Minimal WGSL shaders live in res/ (see res/examples for full scenes).
• Bind group helpers sit in bind_group; pair them with pipeline layouts for material or compute resource binding.
• For GPU-driven draws, see indirect helpers on render_pass::RenderPassInstance.
• The examples under res/examples show complete scenes; start with display_texture for a minimal textured quad.

Re-exports

pub use instance::RenderInstanceData;

pub use passes::command_buffer;

pub use passes::compute_pass;

pub use passes::render_pass;

pub use pipelines::bind_group;

pub use pipelines::compute_pipeline;

pub use pipelines::render_pipeline;

pub use resources::buffer;

pub use resources::texture;

pub use utils::vertex;

Modules

instance

Device/queue/surface bootstrap utilities used across the renderer.

passes

Render and compute pass helpers built on top of wgpu encoders.

pipelines

Pipeline and bind-group construction utilities.

resources

GPU resources shared across the renderer.

utils

Small utilities shared across the renderer.