|Summary:||REGRESSION(r182192): Ligatures do not interact correctly with SHY in some fonts|
|Product:||WebKit||Reporter:||Myles C. Maxfield <mmaxfield>|
|Component:||Text||Assignee:||Myles C. Maxfield <mmaxfield>|
|Version:||WebKit Nightly Build|
Description Myles C. Maxfield 2015-10-10 15:37:08 PDT
Created attachment 262836 [details] Reduction See attached reduction.
Comment 1 Myles C. Maxfield 2015-10-10 15:37:54 PDT
Created attachment 262837 [details] Reduction font
Comment 2 Myles C. Maxfield 2015-10-10 15:41:11 PDT
Occurs on iOS and OS X
Comment 3 Myles C. Maxfield 2015-10-10 15:45:05 PDT
Problem occurs on El Capitan, but does not occur on Yosemite.
Comment 4 Myles C. Maxfield 2015-10-10 16:15:30 PDT
Disregard my last comment. Problem occurs on shipping Safari on Yosemite. On ToT Safari on Yosemite, font-variant-ligatures: common-ligatures; fixes the problem. Because on shipping Safari, the problem goes away when you select the text, this is likely a disagreement between our fast and complex text paths. The previous evidence about font-variant-ligatures seems to agree with this (because font-variant-ligatures forces us onto our complex path)
Comment 5 Myles C. Maxfield 2015-10-10 16:20:20 PDT
(In reply to comment #4) > Disregard my last comment. > > Problem occurs on shipping Safari on Yosemite. > > On ToT Safari on Yosemite, font-variant-ligatures: common-ligatures; fixes > the problem. > > Because on shipping Safari, the problem goes away when you select the text, > this is likely a disagreement between our fast and complex text paths. The > previous evidence about font-variant-ligatures seems to agree with this > (because font-variant-ligatures forces us onto our complex path) Same thing on El Capitan.
Comment 6 Myles C. Maxfield 2015-10-10 16:55:52 PDT
I caused this in r182192. I'm shifting glyphs and advances, but not the rest of GlyphBuffer's contents.
Comment 8 WebKit Commit Bot 2015-10-10 17:12:42 PDT
Attachment 262838 [details] did not pass style-queue: ERROR: Source/WebCore/ChangeLog:15: You should remove the 'No new tests' and either add and list tests, or explain why no new tests were possible. [changelog/nonewtests]  Total errors found: 1 in 3 files If any of these errors are false positives, please file a bug against check-webkit-style.
Comment 11 Myles C. Maxfield 2015-10-10 17:29:44 PDT
Option 2 is preferable.
Comment 12 Alexey Proskuryakov 2015-10-10 17:57:45 PDT
Comment on attachment 262840 [details] Option 2 View in context: https://bugs.webkit.org/attachment.cgi?id=262840&action=review > Source/WebCore/ChangeLog:14 > + However, now that <rdar://problem/20230073> is fixed, we don't need to perform this filter in > + the first place. Please make sure that all Mavericks bots (including EWS) have the appropriate security update installed. I'm pretty sure that they do not, so landing this will make the original problem occur again.