[WebGPU] Update vertex-buffer-triangle-strip.html to actually use vertex buffer
Created attachment 359231 [details] Patch
Comment on attachment 359231 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=359231&action=review > Source/WebCore/platform/graphics/gpu/cocoa/GPURenderPipelineMetal.mm:138 > + // FIXME: Setting bufferIndex to an index that DNE causes a SIGABRT when calling [MTLDevice newRenderPipelineStateWithDescriptor] > + // that is not caught by BLOCK_OBJC_EXCEPTIONS. I don’t think "DNE" is standard terminology. Writing out "does not exist" or "is out of bounds" would be clearer. But this FIXME comment is not very clear. What does the comment mean? It seems to be an interesting bit of trivia; what should we be fixing? It’s like a comment saying "// FIXME: Dereferencing a null pointer causes a SIGABRT that is not caught by BLOCK_OBJC_EXCEPTIONS." True, but how exactly is this relevant to the next line of code? Is it possible that inputSlot is an index that does not exist? If so, what can we do about it?
Comment on attachment 359231 [details] Patch This patch has too many FIXMEs.
Thanks for the comments! Instead of writing in fixmes I'll create a new bug to address the issues I raised for myself.
<rdar://problem/47326063>
Created attachment 359297 [details] Patch
Comment on attachment 359297 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=359297&action=review > Source/WebCore/platform/graphics/gpu/cocoa/GPURenderPipelineMetal.mm:137 > - mtlAttributeDesc.bufferIndex = attributes[i].shaderLocation; > + mtlAttributeDesc.bufferIndex = attributes[i].inputSlot; You don't mention why you made this change. The bug title only mentions the HTML. What was wrong about it? > LayoutTests/webgpu/vertex-buffer-triangle-strip.html:17 > + float4 position [[attribute(0)]]; Can you label with both attribute(0) and position to avoid having two identical structs?
Comment on attachment 359297 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=359297&action=review > LayoutTests/webgpu/vertex-buffer-triangle-strip.html:32 > +fragment float4 fragment_main(VertexOut v [[stage_in]]) Does this need to take any arguments at all?
Created attachment 359336 [details] Patch for landing
Comment on attachment 359336 [details] Patch for landing Clearing flags on attachment: 359336 Committed r240101: <https://trac.webkit.org/changeset/240101>
All reviewed patches have been landed. Closing bug.