834716 - test_resize_move_windows.xul fails on an ubuntu 12.04 VM, causes 45 other failures

Status: RESOLVED FIXED

(Core :: DOM: Core & HTML, defect)

Description

[Joel Maher ( :jmaher ) (UTC -8)]

we are going to be running unittests on EC2 VMs in the near future and test_resize_move_windows.xul (from mochitest-chrome) fails consistently:
https://tbpl.mozilla.org/php/getParsedLog.php?id=19102371

This test times out and then about 45 other failures occur as a result of this failure.  I have verified this by removing this test and running the mochitest-chrome suite and only get a few failures.

If this test requires real hardware, please let us know, otherwise, we need to get this fixed or disabled.

Comment 1

[Mounir Lamouri (:mounir)]

I do not think it requires real hardware. Only real window manager.

Actually, I can reproduce this on my new Fedora. When I wrote the test, I wasn't using the same window manager. I will try to have a look and fix that next week.

This said, we run all our tests in old version of Fedora and I always thought that most of them wouldn't work well in another environment. Is this the only issue you have with running our test suite in a quite recent Ubuntu?

Comment 2

[Joel Maher ( :jmaher ) (UTC -8)]

I am tracking the failures in bug 834725, but there are very few issues as it stands.  I only have 3 other bugs to file.  

thanks for looking at this next week!

Comment 3

[Joel Maher ( :jmaher ) (UTC -8)]

:mounir, any luck with this?

Comment 4

[Mounir Lamouri (:mounir)]

I had a look but I wasn't able to identify the problem. For the moment, it seems that a request to move the window is simply ignored by GTK.

BTW, this test is currently disabled on Linux.

Comment 5

[Joel Maher ( :jmaher ) (UTC -8)]

thanks for taking a look at this last week.  The mochitest-plain test is disabled for linux, but the mochitest-chrome test is still active:
http://dxr.mozilla.org/mozilla-central/dom/tests/mochitest/chrome/Makefile.in.html#l35

Should we disable the mochitest-chrome one for linux as we do for plain?
http://dxr.mozilla.org/mozilla-central/dom/tests/mochitest/bugs/Makefile.in.html#l145

Comment 6

[Karl Tomlinson (:karlt)]

(In reply to Mounir Lamouri (:mounir) from comment #4)
> For the moment, it
> seems that a request to move the window is simply ignored by GTK.

I wouldn't be surprised if the window manager ignored a request to move the window off the screen.

http://hg.mozilla.org/mozilla-central/annotate/de946331d886/dom/tests/mochitest/chrome/test_resize_move_windows.xul#l50

Comment 7

[Joel Maher ( :jmaher ) (UTC -8)]

so what I see happening here is that we resize the window to some really small amount and there is no focus on the window.  If I click the window it moves further along in the test, but it doesn't pass.  I am really thinking we need to disable this for gtk2.

Comment 8

[Joel Maher ( :jmaher ) (UTC -8)]

Attachment: disable the chrome test_resize_windows.xul test for linux (1.0)

Comment 9

[Joel Maher ( :jmaher ) (UTC -8)]

Thanks for the review!

https://hg.mozilla.org/mozilla-central/rev/469333ea459c

Comment 10

[Ed Morley [:emorley]]

Comment on attachment 710268 [details] [diff] [review]
disable the chrome test_resize_windows.xul test for linux (1.0)

>+ifneq (Linux,$(OS_ARCH))
>+MOCHITEST_FILES += \
>+		test_resize_move_windows.xul \
>+		$(NULL)

This should be MOCHITEST_CHROME_FILES (getting timeout on m-c in this test):
https://tbpl.mozilla.org/php/getParsedLog.php?id=19575697&tree=Firefox

Comment 11

[Ed Morley [:emorley]]

Pushed followup:
https://hg.mozilla.org/mozilla-central/rev/fe1600b22c77

(I've very almost made exactly the same mistake - I have a file to copy/paste the makefile disable-on-one-platform boilerplate, and is easy to forget to substitute the test type :-))