.
<rdar://problem/72061985>
Created attachment 438987 [details] Patch
Comment on attachment 438987 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=438987&action=review > Source/WebCore/ChangeLog:16 > + While it is possible to create a dedicated display list item for this, we don't want to do > + that because `PKDrawApplePayButtonWithCornerRadius` involves dealing with PDFs, which are > + not as secure as we'd like for use in the GPUProcess. What about doing something like we do for DrawGlyphsRecorder and use a CGContextDelegate to get the underlying draw calls and serialize those? (This is of course assuming CGContextDelegate doesn't have dedicate DrawPDFPage delegation, but I didn't think it did)?
Comment on attachment 438987 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=438987&action=review >> Source/WebCore/ChangeLog:16 >> + not as secure as we'd like for use in the GPUProcess. > > What about doing something like we do for DrawGlyphsRecorder and use a CGContextDelegate to get the underlying draw calls and serialize those? (This is of course assuming CGContextDelegate doesn't have dedicate DrawPDFPage delegation, but I didn't think it did)? Interesting! I had no idea that existed =D And yeah I dont see anything related to PDF. Frankly though I'm not really sure I have the time to investigate this right now. I'm happy to file a followup to investigate this further if I get a chance in a few weeks. The initial desire for this change is to just get this working in a GPUProcess world, which I think this achieves.
Comment on attachment 438987 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=438987&action=review >>> Source/WebCore/ChangeLog:16 >>> + not as secure as we'd like for use in the GPUProcess. >> >> What about doing something like we do for DrawGlyphsRecorder and use a CGContextDelegate to get the underlying draw calls and serialize those? (This is of course assuming CGContextDelegate doesn't have dedicate DrawPDFPage delegation, but I didn't think it did)? > > Interesting! I had no idea that existed =D And yeah I dont see anything related to PDF. > > Frankly though I'm not really sure I have the time to investigate this right now. I'm happy to file a followup to investigate this further if I get a chance in a few weeks. > > The initial desire for this change is to just get this working in a GPUProcess world, which I think this achieves. We don't have a general purpose recorder, only drawglyphs, so I think this is OK for now (you could imagine us having implemented the whole of GPUP that way instead, but we did not).
Committed r283331 (242352@main): <https://commits.webkit.org/242352@main> All reviewed patches have been landed. Closing bug and clearing flags on attachment 438987 [details].
Please at least make sure there are tests for drawing and scaling these at different sizes to ensure we don't regress that.