Andre Weissflog

...not super critical because sokol_app.h treats the event struct as "write-only", but it's not nice. Nim doesn't seem to have the concept of a const-ptr argument, so it's not an...

Initial (readonly) storage buffer support.

2

First step will introduce storage buffers without compute shaders. This allows more flexibility for providing random-access data to the GPU (e.g. replacing float textures as 'poor man's storage buffers'). -...

Planning: Resource binding cleanup

1

This would make sense after the storage-buffer update: General idea is to allow more flexibility when mapping the sokol-gfx 'bind slot convention' to the backend API bind slots. - Get...

Enable vertex pulling (storage buffers + vertex id)

1

Now that GLES2/WebGL1 has been dropped it's time to look into the topic of "rendering without vertex buffers" again: - don't require a vertex layout specification when creating a pipeline...

This was actually only required because of the 'incompleteness problem' in the GL backend which is fixed by setting GL_TEXTURE_MAX_LEVEL. D3D11 and WebGPU (also Vulkan) don't have the concept of...

Apparently calling `ID3D11Debug::ReportLiveDeviceObjects` reports weird things like query and barrier objects. Investigate!

There's now `_cb` (in sokol_app.h), `log_func` and `alloc_fn`. Unify those all to `_fn` (small but breaking change, probably best to throw together with a bigger breaking change.

Image data for all image attribute combinations should be described by a single ptr/size pair and a clearly defined and documented memory layout. - 2D vs 3D vs Array vs...

This would reduce the memory for keys by half, but required a decoding step during sampling.