Bug 122006 - [wk2] WebKitTestRunner's window snapshots are blank with tiled drawing
Summary: [wk2] WebKitTestRunner's window snapshots are blank with tiled drawing
Status: RESOLVED DUPLICATE of bug 121859
Alias: None
Product: WebKit
Classification: Unclassified
Component: Tools / Tests (show other bugs)
Version: 528+ (Nightly build)
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: Tim Horton
URL:
Keywords: InRadar
Depends on: 122035 122152 122176
Blocks: 121859
  Show dependency treegraph
 
Reported: 2013-09-26 22:34 PDT by Tim Horton
Modified: 2013-10-01 18:41 PDT (History)
6 users (show)

See Also:


Attachments
preliminary (3.81 KB, patch)
2013-09-26 22:52 PDT, Tim Horton
no flags Details | Formatted Diff | Diff
patch (2.35 KB, patch)
2013-09-27 13:35 PDT, Tim Horton
dino: review+
Details | Formatted Diff | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description Tim Horton 2013-09-26 22:34:03 PDT
WebKitTestRunner's window snapshots are pure white no matter what's on screen when they're taken, if tiled drawing is on (possibly any time remote layer hosting is occurring?) and WKTR's window has a buffered backing store.

This doesn't show up in ref tests because it's broken in both the ref and the actual result; pixel tests make it the most clear (but we don't have any tiled-drawing pixel tests, nor do we run the other ones very often).

I've added a mismatch test to ensure this doesn't happen again, as I'm going to switch WKTR to tiled drawing 24/7 very soon, and it would be really bad if we lost all ref image coverage :|

Conveniently, we're really mostly interested in testing unbuffered windows anyway.
Comment 1 Radar WebKit Bug Importer 2013-09-26 22:34:26 PDT
<rdar://problem/15095197>
Comment 2 Tim Horton 2013-09-26 22:52:47 PDT
Created attachment 212789 [details]
preliminary

I'm seeing some tests that use the non-tiled path fail with this change, so it might require more investigation.
Comment 3 Tim Horton 2013-09-27 13:21:44 PDT
We can't override the window resolution with an unbuffered window, so this breaks the workaround we use to make tests work on Retina machines. I'll try to come up with something else.
Comment 4 Tim Horton 2013-09-27 13:35:23 PDT
Created attachment 212837 [details]
patch
Comment 5 Tim Horton 2013-09-27 13:46:11 PDT
http://trac.webkit.org/changeset/156569
Comment 6 WebKit Commit Bot 2013-09-27 14:56:04 PDT
Re-opened since this is blocked by bug 122035
Comment 7 Tim Horton 2013-10-01 11:49:39 PDT
http://trac.webkit.org/changeset/156725
Comment 8 WebKit Commit Bot 2013-10-01 13:33:37 PDT
Re-opened since this is blocked by bug 122176
Comment 9 Tim Horton 2013-10-01 18:41:06 PDT
Meh, I'm just folding this into https://bugs.webkit.org/show_bug.cgi?id=121859

*** This bug has been marked as a duplicate of bug 121859 ***