https://tbpl.mozilla.org/php/getParsedLog.php?id=31608182&tree=Mozilla-Inbound Ubuntu VM 12.04 mozilla-inbound debug test reftest on 2013-12-06 21:17:23 PST for push 84cb9cf72c02 slave: tst-linux32-ec2-402 22:07:06 INFO - REFTEST TEST-START | file:///builds/slave/test/build/tests/reftest/tests/layout/reftests/mathml/mstyle-2.xhtml 22:07:06 INFO - REFTEST TEST-LOAD | file:///builds/slave/test/build/tests/reftest/tests/layout/reftests/mathml/mstyle-2.xhtml | 5718 / 10140 (56%) (a surprising amount of GC) 22:07:06 INFO - REFTEST TEST-LOAD | file:///builds/slave/test/build/tests/reftest/tests/layout/reftests/mathml/mstyle-2-ref.xhtml | 5718 / 10140 (56%) 22:07:06 INFO - --DOMWINDOW == 66 (0xd6fd95c) [pid = 2297] [serial = 15635] [outer = (nil)] [url = file:///builds/slave/test/build/tests/reftest/tests/layout/reftests/mathml/semantics-1.xhtml] 22:07:06 INFO - --DOMWINDOW == 65 (0x9c84b24) [pid = 2297] [serial = 15636] [outer = (nil)] [url = file:///builds/slave/test/build/tests/reftest/tests/layout/reftests/mathml/semantics-1-ref.xhtml] 22:07:06 INFO - ++DOMWINDOW == 66 (0xd5d57fc) [pid = 2297] [serial = 15672] [outer = 0x97e477c] 22:07:07 INFO - REFTEST TEST-UNEXPECTED-FAIL | file:///builds/slave/test/build/tests/reftest/tests/layout/reftests/mathml/mstyle-2.xhtml | image comparison (==), max difference: 255, number of differing pixels: 88 22:07:07 INFO - REFTEST INFO | Saved log: START file:///builds/slave/test/build/tests/reftest/tests/layout/reftests/mathml/mstyle-2.xhtml 22:07:07 INFO - REFTEST INFO | Saved log: [CONTENT] OnDocumentLoad triggering AfterOnLoadScripts 22:07:07 INFO - REFTEST INFO | Saved log: Initializing canvas snapshot 22:07:07 INFO - REFTEST INFO | Saved log: DoDrawWindow 0,0,800,1000 22:07:07 INFO - REFTEST INFO | Saved log: [CONTENT] RecordResult fired 22:07:07 INFO - REFTEST INFO | Saved log: RecordResult fired 22:07:07 INFO - REFTEST INFO | Saved log: START file:///builds/slave/test/build/tests/reftest/tests/layout/reftests/mathml/mstyle-2-ref.xhtml 22:07:07 INFO - REFTEST INFO | Saved log: [CONTENT] OnDocumentLoad triggering AfterOnLoadScripts 22:07:07 INFO - REFTEST INFO | Saved log: Initializing canvas snapshot 22:07:07 INFO - REFTEST INFO | Saved log: DoDrawWindow 0,0,800,1000 22:07:07 INFO - REFTEST INFO | Saved log: [CONTENT] RecordResult fired 22:07:07 INFO - REFTEST INFO | Saved log: RecordResult fired 22:07:07 INFO - REFTEST INFO | Loading a blank page 22:07:07 INFO - ++DOMWINDOW == 67 (0xa87cefc) [pid = 2297] [serial = 15673] [outer = 0x97e477c] 22:07:07 INFO - REFTEST TEST-END | file:///builds/slave/test/build/tests/reftest/tests/layout/reftests/mathml/mstyle-2.xhtml

Perhaps bug 415413 made this test unreliable or is it just coincidence?

As discussed on Karl on IRC, the border is not from a MathML table and is not supposed to be tested here. So a quick fix could be to remove the border. Karl suggested that a wrong preferred width could deceive the table layout. But here the width is 100%, so I'm not sure the preferred width is involved...

I'm aware of the bugs about missing borders for border-collapse tables, but those generally involve column/row-span or zooming. So in this case, I think it might worth investigating if it's a bug in the MathML code. Maybe we could do a Try reftest run (on Linux32 debug) that only runs this test a few thousand times to see if it's something that is actually repeatable at all.

(In reply to Mats Palmgren (:mats) from comment #3) > So in this case, I think it might > worth investigating if it's a bug in the MathML code. Maybe we could do a > Try > reftest run (on Linux32 debug) that only runs this test a few thousand times > to see if it's something that is actually repeatable at all. Could you try again now that bug 838509 is fixed? The mstyle reftests have been updated now that we only support a subset of relevant mstyle attributes mapped to CSS. It would be great to see if this intermittent bug as well as bug 924322 still shows up now...

> Could you try again now that bug 838509 is fixed? I don't see how that's possible to do since this is a very rare intermittent failure. Let's just assume bug 838509 fixed it and resolve it as wfm. It'll be reopened or filed again if it still occurs.

(In reply to Mats Palmgren (:mats) from comment #6) > > Could you try again now that bug 838509 is fixed? > > I don't see how that's possible to do since this is a very rare intermittent > failure. > Let's just assume bug 838509 fixed it and resolve it as wfm. It'll be > reopened or > filed again if it still occurs. OK, I thought you were able "do a Try reftest run (on Linux32 debug) that only runs this test a few thousand times", but I'm fine with closing the bug.