NEW 160120
[GTK] Unskip layout test editing/deleting/delete-emoji.html
https://bugs.webkit.org/show_bug.cgi?id=160120
Summary [GTK] Unskip layout test editing/deleting/delete-emoji.html
Michael Catanzaro
Reported 2016-07-23 09:41:21 PDT
I'm skipping the layout test editing/deleting/delete-emoji.html because it doesn't have expected results yet, and the actual results I see do not look reasonable. On Mac there are a bunch of emoji there, but on GTK there are only placeholder characters shown. Something in the GTK stack doesn't support these new emoji properly.
Attachments
Bastien Nocera
Comment 1 2016-10-11 21:11:13 PDT
Duplicate of 156579?
Michael Catanzaro
Comment 2 2016-10-12 09:33:08 PDT
I'm not sure. I would assume bug #156579 would be just about adding color to the emoji. The emoji in this test are not drawn properly at all.
Adrian Perez
Comment 3 2017-10-24 12:09:33 PDT
(In reply to Michael Catanzaro from comment #0) > I'm skipping the layout test editing/deleting/delete-emoji.html because it > doesn't have expected results yet, and the actual results I see do not look > reasonable. On Mac there are a bunch of emoji there, but on GTK there are > only placeholder characters shown. Something in the GTK stack doesn't > support these new emoji properly. Today I have been able to verify that the output for the GTK+ port matches that of the Mac port for the following tests: * editing/deleting/delete-emoji.html * editing/deleting/delete-emoji-1.html I'll add the expectations and unskip these two only, and check the rest of the emoji deletion tests later on.
Adrian Perez
Comment 4 2017-10-25 01:05:50 PDT
(In reply to Adrian Perez from comment #3) > [...] > * editing/deleting/delete-emoji.html > * editing/deleting/delete-emoji-1.html > > I'll add the expectations and unskip these two only, and check the rest of > the emoji deletion tests later on. Done in bug #178742
Myles C. Maxfield
Comment 5 2017-10-26 11:43:09 PDT
https://trac.webkit.org/changeset/223578/webkit This is the patch which caused this. This patch removes custom character-iteration code inside RenderText in favor of platform-supplied character iterators. On GTK, this would take the form of ubrk_ ICU functions using the CHARACTER enum. The best thing to do would be to verify that the deletion behavior matches some other native app on GTK, and if so, generate expected results.
Note You need to log in before you can comment on or make changes to this bug.