Module stream_buffer 

Persistent per-pipeline streaming buffers.

Each frame, the renderer accumulates vertex data into per-pipeline batches and flushes them to the GPU. The historical code path called device.create_buffer_init on every flush, allocating and dropping a fresh wgpu::Buffer pair (vertex + index) per batch — a known antipattern that causes driver allocation pressure at high glyph counts.

StreamBuffer owns a single growable GPU buffer that persists across frames. StreamBuffer::ensure_capacity is called once per frame with the worst-case byte count (requires &mut self), growing the underlying buffer if needed. After that, StreamBuffer::write copies a batch to the GPU at the current write offset and advances the cursor — all via interior mutability, so calls can interleave freely with other &self methods on the renderer.

Structs

StreamBuffer

A growable, ring-reset GPU buffer for streaming vertex or index data across contiguous batches within a single frame.

StreamBuffers

All per-pipeline streaming buffers owned by a Renderer.