875331 - tabs.closeAllTabs() makes waitForPageLoad() to fail in 2.0

Status: RESOLVED WORKSFORME

(Mozilla QA Graveyard :: Mozmill Tests, defect)

Description

[Andreea Matei [:AndreeaMatei]]

I discovered this while working on bug 758187, there we close all tabs in setupModule(), before the test itself. The failure I got was regarding waitForPageLoad() but I could see the page being loaded though.

Adding a testcase to reproduce, without using the libraries. If you remove the code that close all tabs from setup, test works.

This is reproducing on Linux, all branches except Aurora.

Comment 1

[Andreea Matei [:AndreeaMatei]]

Attachment: testcase

Comment 2

[Henrik Skupin [:whimboo][⌚️UTC+2] (away 10/03 - 10/13)]

This testcase is passing for me for nightly, aurora, and beta. Can you still reproduce? If yes, I would like to see the full debug output.

Comment 3

[Andreea Matei [:AndreeaMatei]]

Attachment: failing beta

This is the debugging log with the dumps from windows.js and some in the testcase code.

Comment 4

[Andreea Matei [:AndreeaMatei]]

Attachment: passing beta without closeAllTabs()

This is passing without the closeAllTabs() code.

Tests were made on Linux 12.04, 64 bit, latest beta.

Comment 5

[Henrik Skupin [:whimboo][⌚️UTC+2] (away 10/03 - 10/13)]

(In reply to Andreea Matei [:AndreeaMatei] from comment #3)
> Created attachment 754698 [details]
> failing beta

> ** Page has been finished loading: id=13, status=true, uuid={936d710e-91a1-4479-bfc8-79320751e670}
> TEST-START | /home/andreeamatei/geolocation/tests/functional/testGeolocation/testcase.js | testGeolocation
> ** Page has been finished loading: id=13, status=false, uuid={936d710e-91a1-4479-bfc8-79320751e670}
> * current map: {"3":{"loaded":true},"13":{"loaded":true,"id_load_in_transition":{"name":"","number":"{936d710e-91a1-4479-bfc8-79320751e670}","valid":true},"id_load_handled":{"name":"","number":"{936d710e-91a1-4479-bfc8-79320751e670}","valid":true}}}

It looks that we call waitForPageLoad twice. In line 1 you can see that the status is true and we update the internal map accordingly with the new uuid. Then in line 3 we started again to wait for the page being loaded, cause status is false. Where does it come from?

Comment 6

[Andreea Matei [:AndreeaMatei]]

Hm, could be from the 2 tabs from the start? I mean, we close 1 tab, then the other one comes in focus, so it might wait for it to load. Then we open our geolocation page. I'll have to check this.

Comment 7

[Henrik Skupin [:whimboo][⌚️UTC+2] (away 10/03 - 10/13)]

Has anything been changed here given that we make use of the new tab page now?

Comment 8

[Andreea Matei [:AndreeaMatei]]

Heh, indeed it's no longer reproducing, was fixed by bug 764782. Yey!

Comment 9

[Henrik Skupin [:whimboo][⌚️UTC+2] (away 10/03 - 10/13)]

Well, that doesn't mean it is fixed. We should still not fail when we try to open a new tab and directly load about:blank.

Comment 10

[Andreea Matei [:AndreeaMatei]]

I used the current testcase added here, which passes now (and it's with the old closeAllTabs() code that uses about:blank).

Now checked with about:newtab and also passes.
I think it was fixed by another change?

Here's the output with about:newtab :
http://pastebin.mozilla.org/2523439

And here with about:blank :
http://pastebin.mozilla.org/2523445

Comment 11

[Henrik Skupin [:whimboo][⌚️UTC+2] (away 10/03 - 10/13)]

Ok, good to hear that. So yes, lets close it out.