Closed Bug 1023669 Opened 12 years ago Closed 11 years ago

Intermittent variable-supports-01.html,variable-supports-41.html,variable-supports-54.html,variable-supports-55.html | image comparison (==), max difference: 255, number of differing pixels: 800000

Categories

(Core :: CSS Parsing and Computation, defect)

Product:
Core
Component:
CSS Parsing and Computation
Platform:
x86_64
Windows 7
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

()

Status:
RESOLVED WORKSFORME

People

(Reporter: KWierso, Unassigned)

References

Details

(Keywords: intermittent-failure)

https://tbpl.mozilla.org/php/getParsedLog.php?id=41476067&tree=Fx-Team Windows 7 32-bit fx-team pgo test reftest-no-accel on 2014-06-10 15:02:27 PDT for push d93b579522fd slave: t-w732-ix-037 15:06:55 INFO - REFTEST TEST-START | file:///C:/slave/test/build/tests/reftest/tests/layout/reftests/w3c-css/submitted/variables/variable-supports-55.html 15:06:55 INFO - REFTEST TEST-LOAD | file:///C:/slave/test/build/tests/reftest/tests/layout/reftests/w3c-css/submitted/variables/variable-supports-55.html | 1064 / 10985 (9%) 15:06:55 INFO - REFTEST TEST-LOAD | file:///C:/slave/test/build/tests/reftest/tests/layout/reftests/w3c-css/submitted/variables/support/color-green-ref.html | 1064 / 10985 (9%) 15:06:55 INFO - REFTEST TEST-UNEXPECTED-FAIL | file:///C:/slave/test/build/tests/reftest/tests/layout/reftests/w3c-css/submitted/variables/variable-supports-55.html | image comparison (==), max difference: 255, number of differing pixels: 800000 15:06:56 INFO - REFTEST INFO | Saved log: START file:///C:/slave/test/build/tests/reftest/tests/layout/reftests/w3c-css/submitted/variables/variable-supports-55.html 15:06:56 INFO - REFTEST INFO | Saved log: [CONTENT] OnDocumentLoad triggering AfterOnLoadScripts 15:06:56 INFO - REFTEST INFO | Saved log: Initializing canvas snapshot 15:06:56 INFO - REFTEST INFO | Saved log: DoDrawWindow 0,0,800,1000 15:06:56 INFO - REFTEST INFO | Saved log: [CONTENT] RecordResult fired 15:06:56 INFO - REFTEST INFO | Saved log: RecordResult fired 15:06:56 INFO - REFTEST INFO | Saved log: START file:///C:/slave/test/build/tests/reftest/tests/layout/reftests/w3c-css/submitted/variables/support/color-green-ref.html 15:06:56 INFO - REFTEST INFO | Saved log: [CONTENT] OnDocumentLoad triggering AfterOnLoadScripts 15:06:56 INFO - REFTEST INFO | Saved log: Initializing canvas snapshot 15:06:56 INFO - REFTEST INFO | Saved log: DoDrawWindow 0,0,800,1000 15:06:56 INFO - REFTEST INFO | Saved log: [CONTENT] RecordResult fired 15:06:56 INFO - REFTEST INFO | Saved log: RecordResult fired 15:06:56 INFO - REFTEST INFO | Loading a blank page 15:06:56 INFO - REFTEST TEST-END | file:///C:/slave/test/build/tests/reftest/tests/layout/reftests/w3c-css/submitted/variables/variable-supports-55.html
Cameron, see the dupes I'm about to mark. This appears to be somewhat widespread on Win7 Ru jobs. Any ideas what might be going on? I attached screenshots to bug 1036043, but basically its "test loads, reference doesn't"
Flags: needinfo?(cam)
Summary: Intermittent variable-supports-55.html | image comparison (==), max difference: 255, number of differing pixels: 800000 → Intermittent variable-supports-01.html,variable-supports-41.html,variable-supports-54.html,variable-supports-55.html | image comparison (==), max difference: 255, number of differing pixels: 800000
I'm not really sure what's going on. In this bug, looks like it's the test that is not painting, but the reference is, while it's the other way around for bug 1036043. Oddly, the snapshot for the file that doesn't paint is a purely transparent image. If it was just something wrong with the content in the test not painting, I would expect a solid white background. David do you know how it might be that the reftest harness gets a blank snapshot like this, especially given we're passing "rgb(255,255,255)" into the drawWindow call in reftest.js? There's an early return in CanvasRenderingContext2D::DrawWindow (if the target canvas has a scale-to-zero transform on it) and in PresShell::RenderDocument (if we're in the process of being restored). Seems unlikely we'd hit these cases though? Ryan I take it you haven't noticed this outside of the CSS Variables reftests?
Flags: needinfo?(cam) → needinfo?(dbaron)
We've seen similar "image shows in one, no image in the other" types of issues before, but filing bug 1036043 today is when I noticed the seeming pattern that seems to be hitting these tests in particular.
(In reply to Cameron McCormack (:heycam) (travelling until Aug 31) from comment #5) > I'm not really sure what's going on. In this bug, looks like it's the test > that is not painting, but the reference is, while it's the other way around > for bug 1036043. Oddly, the snapshot for the file that doesn't paint is a > purely transparent image. If it was just something wrong with the content > in the test not painting, I would expect a solid white background. David do > you know how it might be that the reftest harness gets a blank snapshot like > this, especially given we're passing "rgb(255,255,255)" into the drawWindow > call in reftest.js? There's an early return in > CanvasRenderingContext2D::DrawWindow (if the target canvas has a > scale-to-zero transform on it) and in PresShell::RenderDocument (if we're in > the process of being restored). Seems unlikely we'd hit these cases though? Yeah, I really don't have any ideas on how this could happen. It shouldn't.
Flags: needinfo?(dbaron)
Inactive; closing (see bug 1180138).
Status: NEW → RESOLVED
Closed: 11 years ago
Resolution: --- → WORKSFORME
You need to log in before you can comment on or make changes to this bug.