Closed MarkCallow closed 5 months ago
@fluppeteer, please review. I was unable to add your GitHub id to the reviewers list.
I think this mostly looks good, except that the multi-channel packing convention wires got crossed. I've added notes where I think the numbers should be different. Please poke me if you change them and want a re-review, or want to push back; I'll try to respond promptly this time.
The #define approach looks much easier to read (not a phrase I'll often say) - thanks for clarifying, and sorry again to have caused the churn!
Assuming all the test are happy, I think this is good to go, and we can talk about coordinate offsets separately.
Code is believed complete. Tests have yet to be added. Hence draft status.
This includes a new round-trip test of vk2dfd and dfd2vk for all Vulkan formats except multi-plane and modifications to both functions, or rather their generators, to fix support for some formats, including VK_FORMAT_R16G16_S10_5_NV and the YUV 422 single plane formats. Those listed required changes to
interpretDFD
.Fixes #859.