Bug 202545

Summary: Fix WebGL 1 conformance regressions when USE_ANGLE=1
Product: WebKit Reporter: James Darpinian <jdarpinian>
Component: ANGLEAssignee: James Darpinian <jdarpinian>
Status: RESOLVED FIXED    
Severity: Normal CC: achristensen, cdumez, commit-queue, darin, dino, esprehn+autocc, ews-watchlist, graouts, gyuyoung.kim, kbr, kondapallykalyan, webkit-bug-importer
Priority: P2 Keywords: InRadar
Version: WebKit Nightly Build   
Hardware: Unspecified   
OS: Unspecified   
Bug Depends on:    
Bug Blocks: 198948    
Attachments:
Description Flags
Fix WebGL 1 conformance regressions when USE_ANGLE=1 none

James Darpinian
Reported 2019-10-03 15:17:23 PDT
Fix WebGL 1 conformance regressions when USE_ANGLE=1
Attachments
Fix WebGL 1 conformance regressions when USE_ANGLE=1 (6.93 KB, patch)
2019-10-03 15:19 PDT, James Darpinian 		no flags
DetailsFormatted DiffDiff
View All Add attachment proposed patch, testcase, etc.
James Darpinian
Comment 1 2019-10-03 15:19:52 PDT
Created attachment 380163 [details] Fix WebGL 1 conformance regressions when USE_ANGLE=1
James Darpinian
Comment 2 2019-10-03 15:23:33 PDT
After this patch plus a couple of upstream changes I'm working on in ANGLE, setting USE_ANGLE=1 on Mac fixes a bunch of WebGL conformance tests and causes no new test regressions.
WebKit Commit Bot
Comment 3 2019-10-03 20:48:18 PDT
Comment on attachment 380163 [details] Fix WebGL 1 conformance regressions when USE_ANGLE=1 Clearing flags on attachment: 380163 Committed r250697: <https://trac.webkit.org/changeset/250697>
WebKit Commit Bot
Comment 4 2019-10-03 20:48:19 PDT
All reviewed patches have been landed. Closing bug.
Radar WebKit Bug Importer
Comment 5 2019-10-03 20:49:16 PDT
<rdar://problem/55972310>
Darin Adler
Comment 6 2019-10-08 09:32:04 PDT
Comment on attachment 380163 [details] Fix WebGL 1 conformance regressions when USE_ANGLE=1 View in context: https://bugs.webkit.org/attachment.cgi?id=380163&action=review > Source/WebCore/html/canvas/WebGLDebugShaders.cpp:54 > if (m_context.isContextLost()) > return String(); Since this function can no longer return null, it doesn’t make sense to return a null string here. Before the change to the bindings, this would turn into a JavaScript null, so this line of code may have had a valuable effect. We should consider returning an empty string here. Or perhaps removing this code entirely as long as the validateWebGLObject function is guaranteed to return false when the context is lost.
Note You need to log in before you can comment on or make changes to this bug.