1390118 - TEST-UNEXPECTED-FAIL | toolkit/components/contextualidentity/tests/unit/test_basic.js | xpcshell return code: 0

Status: RESOLVED FIXED

(Thunderbird :: General, defect)

Description

[Jorg K (CEST = GMT+2)]

TEST-UNEXPECTED-FAIL | toolkit/components/contextualidentity/tests/unit/test_basic.js | xpcshell return code: 0

First seen Mon Aug 14, 2017, 12:11:19
https://treeherder.mozilla.org/#/jobs?repo=comm-central&revision=0710b9ff48cd9419fbf3b7275f84a48c94f71860

M-C last good: 3bfcbdf5c6c381d5a8febb5c209e27a69f
M-C first bad: df9beb781895fcd0493c21e95ad313e004

https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=3bfcbdf5c6c381d5a8febb5c209e27a69f&tochange=df9beb781895fcd0493c21e95ad313e004

Suspect bug 1344519 since that changed test_basic.js:
https://hg.mozilla.org/mozilla-central/rev/3323bbf5cea8

The log says:
04:53:04     INFO -  Unexpected exception NS_ERROR_FILE_NOT_FOUND: Component returned failure code: 0x80520012 (NS_ERROR_FILE_NOT_FOUND) [nsIStringBundle.GetStringFromName]
04:53:04     INFO -  getUserContextLabel@resource://gre/modules/ContextualIdentityService.jsm:330:12

And line 330: reads:
    return gBrowserBundle.GetStringFromName(identity.l10nID);

To me that looks like an invalid dependency between toolkit and browser, so the test fill fail in any context but browser.

Perhaps you therefore added those strings for android:
https://hg.mozilla.org/mozilla-central/rev/3323bbf5cea8#l1.12

Jonathan and Andrew: Sorry to trouble you, but perhaps you can fix this.

Comment 1

[Jonathan Kingston [:jkt] he/him]

I think we should disable the test for anything that is failing that isn't desktop for the time being. I'm not sure why this seems to be intermittent either as it passed through try.

Comment 2

[Jonathan Kingston [:jkt] he/him]

Could it be that the file isn't accessible in tests only? Nothing actually has changed for this file just that it is accessed a little more frequently?

Comment 3

[Jorg K (CEST = GMT+2)]

OK, looks like you're going to work on it, thanks. Please change product/component/summary accordingly.
We usually add
  skip-if = appname == "thunderbird"
for the tests we want to disable in M-C.

Comment 4

[Jonathan Kingston [:jkt] he/him]

Oh wait is this just thunderbird, I couldn't try from the try?
I see the component now which makes more sense about the failure.
Thanks

Comment 5

[Jorg K (CEST = GMT+2)]

Umm, not sure what info you need. Yes, I reported that this test fails when run on the C-C treeherder:
https://treeherder.mozilla.org/#/jobs?repo=comm-central&revision=0710b9ff48cd9419fbf3b7275f84a48c94f71860&selectedJob=122966816

Usually I send an M-C patch to disable test which don't work and can't be made to work.

However, this test would most likely work in TB, if it had the browser strings that are used in the test. So as I said in comment #0, you've created a dependency between toolkit and browser. If you moved the strings out of browser the test would most likely pass for us.

Comment 6

[Jonathan Kingston [:jkt] he/him]

I actually didn't make that dependency, I just exposed the android issue by increasing test exposure. We could move the strings I suspect however would Containers ever make any sense to thunderbird? My hunch is not at all, at least without a tonne of work anyway.

I will patch up ini file to skip the test for the time being.

I appreciate your help on this.

Comment 7

[Jonathan Kingston [:jkt] he/him]

Attachment: Bug 1390118 - Skip contextual identity test for Thunderbird

Review commit: https://reviewboard.mozilla.org/r/168280/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/168280/

Comment 8

[Jonathan Kingston [:jkt] he/him]

I wasn't able to push to try as I got the following error: "warning: repository is unrelated"
I'm not sure if you are able to test?

Comment 9

[Jorg K (CEST = GMT+2)]

Comment on attachment 8896976 [details]
Bug 1390118 - Skip contextual identity test for Thunderbird

Thanks, but I really don't have any review authority in toolkit/ being a Thunderbird and Mailnews peer. Anyway, the change looks good and won't break FF ;-)

Comment 10

[Jorg K (CEST = GMT+2)]

Comment on attachment 8896976 [details]
Bug 1390118 - Skip contextual identity test for Thunderbird

https://reviewboard.mozilla.org/r/168278/#review173470

Thanks, but I really don't have any review authority in toolkit/ being a Thunderbird and Mailnews peer. Anyway, the change looks good and won't break FF ;-)

Comment 11

[Jorg K (CEST = GMT+2)]

(In reply to Jonathan Kingston [:jkt] from comment #8)
> I wasn't able to push to try as I got the following error: "warning:
> repository is unrelated"
> I'm not sure if you are able to test?
Yes, I can ship this off to a C-C try, but all that will happen is that the test won't and and won't fail.

Comment 12

[Jorg K (CEST = GMT+2)]

Comment on attachment 8896976 [details]
Bug 1390118 - Skip contextual identity test for Thunderbird

https://reviewboard.mozilla.org/r/168280/#review173472

Comment 13

[Andrew Swan [:aswan]]

Comment on attachment 8896976 [details]
Bug 1390118 - Skip contextual identity test for Thunderbird

https://reviewboard.mozilla.org/r/168280/#review173476

Comment 14

[Jorg K (CEST = GMT+2)]

Ryan, can you please land this for me.

Comment 15

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/ea391b57ec4e
Skip contextual identity test for Thunderbird r=aswan,jorgk+176914

Comment 16

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/ea391b57ec4e