486787 - New Reftest font-face/src-list-local-localized.html intermittently |timed out waiting for onload to fire|, (and always on MacOSX (10.4) SeaMonkey)

Status: RESOLVED WONTFIX

(Core :: Graphics, defect)

Description

[Serge Gautherie (:sgautherie)]

{
http://tinderbox.mozilla.org/showlog.cgi?log=SeaMonkey/1238743680.1238749649.16715.gz
OS X 10.4 comm-central unit test on 2009/04/03 00:28:00

REFTEST TEST-UNEXPECTED-FAIL | http://localhost:4444/36/font-face/src-list-local-localized.html | timed out waiting for onload to fire
}

Regression timeframe:
http://hg.mozilla.org/releases/mozilla-1.9.1/pushloghtml?startdate=2009-04-02+19%3A00%3A04&enddate=2009-04-03+00%3A25%3A01
only and obvious bug 468387

Comment 1

[Serge Gautherie (:sgautherie)]

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1238930310.1238941465.28284.gz
Linux mozilla-central unit test on 2009/04/05 04:18:30

Comment 2

[Serge Gautherie (:sgautherie)]

Do we need a workaround like
http://mxr.mozilla.org/mozilla-central/source/uriloader/exthandler/tests/unit/test_punycodeURIs.js#83
?

Comment 3

[Robert Kaiser]

I thought reftests have some better way of working around this than xpcshell ones?

jdaggett, can we do something about this happening on Tiger? Is there a real bug we still hit on Tiger here?

Comment 4

[John Daggett (:jtd)]

My guess is that this is just too slow for the 10.4 machine that's hitting this problem.  The solution is to break the test out into several chunks and load only a subset per page.

Comment 5

[Serge Gautherie (:sgautherie)]

(In reply to comment #4)
> My guess is that this is just too slow for the 10.4 machine

Ah, could be!

> The solution is to break the test out into several chunks and load
> only a subset per page.

Ftr, internal reftest timeout is 10 seconds.

Other options would be to either:
*Increase that timeout :-|
*Allow to override it on a per-test basis :-)

David, what do you think ?

Comment 6

[Robert Kaiser]

(In reply to comment #4)
> My guess is that this is just too slow for the 10.4 machine that's hitting this
> problem.

That's possible, this is a mini, which isn't always the fastest box to run things on ;-)

> The solution is to break the test out into several chunks and load
> only a subset per page.

Sounds like a good idea, and probably also increases test granularity, right?

Comment 7

[Serge Gautherie (:sgautherie)]

(In reply to comment #1)
> http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1238930310.1238941465.28284.gz
> Linux mozilla-central unit test on 2009/04/05 04:18:30

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1239505494.1239513008.29699.gz
Linux mozilla-central unit test on 2009/04/11 20:04:54

Comment 8

[Serge Gautherie (:sgautherie)]

KaiRo, could you check how long this test currently takes to run?
As a hint to how many chunks are needed.

Comment 9

[John Daggett (:jtd)]

Skip this test until timing problems worked out
http://hg.mozilla.org/mozilla-central/rev/756dd46daf6c
http://hg.mozilla.org/releases/mozilla-1.9.1/rev/604650c45e71

Comment 10

[Robert Kaiser]

(In reply to comment #8)
> KaiRo, could you check how long this test currently takes to run?

I have no idea how to do that, sorry.

Comment 11

[Serge Gautherie (:sgautherie)]

1. Increase |const LOAD_FAILURE_TIMEOUT = 10000; // ms| in <srcdir\mozilla\layout\tools\reftest\reftest.js>.
2. Execute |python objdir/mozilla/_tests/reftest/runreftest.py srcdir/mozilla/layout/reftests/font-face/reftest.list|.
3. The |REFTEST FINISHED: Slowest test took ...| line should tell us.

Comment 12

[Robert Kaiser]

Hrm, that needs manual interaction on a machine that's supposed to run exclusively automatically. And it needs me to log in via VNC. Lots of work on a day when I want to reduce work a lot currently :(

Comment 13

[Samuel Sidler (old account; do not CC)]

Skipping the test (comment 9) is good enough for 1.9.1.x. If you want a more complete fix, please request approval on an appropriate patch.

Comment 14

[Joe Drew (not getting mail)]

John tells me this is just a timeout from getting slow VMs, so we can't really block on this.