| Summary: | [Qt] QtWebKit -webkit-transform:translate3d fails on big DIV elements | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Product: | WebKit | Reporter: | Jonni Rainisto <jonni.rainisto> | ||||||
| Component: | Layout and Rendering | Assignee: | Nobody <webkit-unassigned> | ||||||
| Status: | RESOLVED WORKSFORME | ||||||||
| Severity: | Normal | CC: | ademar, benjamin, diegohcg, hausmann, laszlo.gombos, noam, suresh.voruganti | ||||||
| Priority: | P2 | Keywords: | Qt, QtTriaged | ||||||
| Version: | 528+ (Nightly build) | ||||||||
| Hardware: | All | ||||||||
| OS: | All | ||||||||
| Bug Depends on: | 39691 | ||||||||
| Bug Blocks: | |||||||||
| Attachments: |
|
||||||||
Cannot reproduce this bug. Please check if it is still reproducible for you. If so try to reproduce it in a open platform. I can still reproduce this in Harmattan week31 release, so its still valid. Seems to be releated to QTWebKit opengl es2 backed, since div stays red if I do --graphicssystem=raster -software. Cannot reproduce on open platform (as I dont have any open hw which has gles2 backend). I can still reproduce this in Harmattan 38-8 QtWebKit using webwidgetrunner (red div disapears when you try to scroll the screen. Is this still reproducible? Can you give Meego Bugzilla ID? Yes, its still reproducible in 43-6. I created now also PMO bug in https://projects.maemo.org/bugzilla/show_bug.cgi?id=200690 Could this be a problem in the OpenGL paint engine? I think MAX_TEXTURE_SIZE is 2048 in this case. Textures bigger than that just fail. This could be worked around by adding tiling for texture, either in WebKit or (better) in the graphics system. I am tempted to just ignore that for the graphics view implementation and do a proper fix for the texture mapper. How important is the bug? (In reply to comment #7) > How important is the bug? It originates from a Maemo 6 bug (41137), triggered by the Qt Web Runtime. See https://bugs.webkit.org/show_bug.cgi?id=39691, I think it fixes it, though the patch needs to be re-applied etc. Please confirm if bug 39691 fixes it so that I can backport/cherry-pick the fix into qtwebkit-2.1 I think we cannot test with out the fix of 39691 integrated to Qtwebkit 2.1 Let me know if we can test with out the fix being picked to 2.1? Created attachment 72856 [details]
patch enabling tiling on qtwebkit-2.1
Patch backported to qtwebkit-2.1. Please test, if it really solves the problem I'll push it into the official branch.
applied the patch to harmattan 43-6, and tested it, but bug still appears. So this patch does not help.(In reply to comment #12) > Created an attachment (id=72856) [details] > patch enabling tiling on qtwebkit-2.1 > > Patch backported to qtwebkit-2.1. Please test, if it really solves the problem I'll push it into the official branch. applied the patch to harmattan 43-6 qtwebkit sources, and tested it, but bug still appears. So this patch does not seem to help. Can we work around this for now (e.g. by using smaller layers in HTML) and postpone actual fix to 2.2? |
Created attachment 59623 [details] scrolling-testcase Following scrolling-testcase.html file fails in Harmattan QtWebkit (red div turns black/disapears), and works on other platforms. Maybe QtWebKit OpenGL backend releated bug?