It was hardly broken locally so I took some time to fix what I was seeing. It seems to work fine now so I think we should give re-enabling a try. I did two try runs: https://tbpl.mozilla.org/?tree=Try&rev=25dde0a7eba7 https://tbpl.mozilla.org/?tree=Try&rev=4f97dcb62f9d The first one has M-3 oranges that are fixed in the second one except the B2G ones which I fixed by disabling the test on B2G. Actually, the test can't run on B2G given that it consist of changing the size and the position of a window. That's not doable on our mobile implementations (Fx Android and Fx OS).

Comment on attachment 728641 [details] [diff] [review] Patch Haven't actually reviewed, but more test coverage is better, so rs=me.

Backed out: https://hg.mozilla.org/integration/mozilla-inbound/rev/6a5d309283df

It did manage to pass a couple of times, which is nice, but when it doesn't, most of the time, it leaves the browser window shrunken down to the smallest possible size, which makes pretty much every test rather angry, and winds up taking two hours instead of twenty minutes.

Why has this been backed out? I do not see any reason given in the bug? Did tests failed? If yes, which tests? Can I have a log? Bug 677841 wasn't updated so I'm assuming no new occurrence of it happened.

Thanks for the link. So it seems that one of the two tests was completely fine. I re-landed the patch by only re-enabling the non-chrome test. The issue I believe is that our old Fedora and the new Ubuntu systems don't use the same gtk versions so things can hardly work well in both (given that the tests seem to fail because gtk refuses to do some changes). https://hg.mozilla.org/integration/mozilla-inbound/rev/a7e1bb7d0e21 Benjamin, sorry for yelling at you on IRC. However, I was really missing the information I needed to understand what happened.

You're right. My apologies for wasting your time.

Comment on attachment 728641 [details] [diff] [review] Patch Review of attachment 728641 [details] [diff] [review]: ----------------------------------------------------------------- thanks for fixing this. Does this look good on try server? ::: dom/tests/mochitest/chrome/Makefile.in @@ -74,5 @@ > - $(NULL) > -else > -$(filter disabled-on-linux-for-timeouts--bug-834716, test_resize_move_windows.xul) > -endif > - In general, I don't like this. We are working towards moving into a manifest based solution for which tests run and don't run on a given platform or configuration.