> New renderers are only required for new graphics languages, which are quite rare.
Two examples:
* DirectFB works with its own image/surface types. The information provided by a LinearBuffer (starting address, width, height, stride etc) is not sufficient.
* Similarly 2D blitters may need more or different information than a LinearBuffer provides, thus requiring a new buffer and renderer type.
But even if we accept that GL/Vulkan/Linear are enough, what's the benefit of the core knowing about these types?
> New renderers are only required for new graphics languages, which are quite rare.
Two examples:
* DirectFB works with its own image/surface types. The information provided by a LinearBuffer (starting address, width, height, stride etc) is not sufficient.
* Similarly 2D blitters may need more or different information than a LinearBuffer provides, thus requiring a new buffer and renderer type.
But even if we accept that GL/Vulkan/Linear are enough, what's the benefit of the core knowing about these types?