Stop gap patch fix for regression in r267329.
Created attachment 411980 [details] Patch
Created attachment 412012 [details] Patch
Comment on attachment 412012 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=412012&action=review > Source/WebCore/ChangeLog:7 > + What was the regression? How does this fix it? Where is the long term solution being tracked?
Created attachment 412017 [details] Patch
Created attachment 412019 [details] Patch
Comment on attachment 412019 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=412019&action=review Looks good. I really appreciate having the test, which I will keep working and/or refine as I return to the Selection API work. > Source/WebCore/editing/VisibleSelection.cpp:390 > - bool shouldUpdateAnchor = m_start != startBeforeAdjustments; > - bool shouldUpdateFocus = m_end != endBeforeAdjustments; > + bool shouldUpdateAnchor = false; > + bool shouldUpdateFocus = false; Might be good to leave a comment behind here. The fact that these are always false makes the code peculiar for now, but makes sense if you know we intend to change it later.
Created attachment 412033 [details] Patch
Created attachment 412039 [details] Patch
Comment on attachment 412039 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=412039&action=review > Source/WebCore/editing/VisibleSelection.cpp:389 > + bool shouldUpdateAnchor = false; // Set to false because of <rdar://problem/69542459>. Can be returned to original logic when this problem is fully fixed. Probably more kind to open source contributors to cite https://bugs.webkit.org/show_bug.cgi?id=218020 instead of rdar, but clear enough either way. I promise to remove this comment some day either way.
Committed r268847: <https://trac.webkit.org/changeset/268847> All reviewed patches have been landed. Closing bug and clearing flags on attachment 412039 [details].
<rdar://problem/70553518>
Seems like the test isn’t working well to detect the problem. I rolled the change back and the test still passes!
No, I was running the test wrong. I was confused by the ChangeLog, which lists the wrong directory for the test.