Closed Bug 1167986 Opened 4 years ago Closed 4 years ago

TEST-UNEXPECTED-FAIL | xpcshell-unpack.ini:toolkit/mozapps/extensions/test/xpcshell/test_signed_install.js (and signing failures)

Categories

(Toolkit :: Add-ons Manager, defect)

defect
Not set

Tracking

()

RESOLVED FIXED
mozilla43
Tracking Status
firefox41 --- fixed
firefox42 --- fixed
firefox43 --- fixed
firefox-esr38 --- unaffected

People

(Reporter: mkmelin, Assigned: Fallen)

References

Details

(Keywords: intermittent-failure, regression)

Attachments

(1 file, 3 obsolete files)

Timing is a little too convenient for this not to be from bug 1164168.

05:09:44  WARNING -  TEST-UNEXPECTED-FAIL | xpcshell-unpack.ini:toolkit/mozapps/extensions/test/xpcshell/test_signed_install.js |  - 6 == 4
05:09:44     INFO -      /builds/slave/test/build/tests/xpcshell/tests/toolkit/mozapps/extensions/test/xpcshell/test_signed_install.js:test_install_broken:110
05:09:44     INFO -      self-hosted:next:624
05:09:44     INFO -      _run_next_test@/builds/slave/test/build/tests/xpcshell/head.js:1440:9
05:09:44     INFO -      do_execute_soon/<.run@/builds/slave/test/build/tests/xpcshell/head.js:653:9
05:09:44     INFO -      _do_main@/builds/slave/test/build/tests/xpcshell/head.js:207:5
05:09:44     INFO -      _execute_test@/builds/slave/test/build/tests/xpcshell/head.js:513:5
05:09:44     INFO -      @-e:1:1
05:09:44     INFO -  exiting test
05:09:44     INFO -  Unexpected exception 2147500036

http://ftp.mozilla.org/pub/mozilla.org/thunderbird/nightly/2015/05/2015-05-24-03-02-20-comm-central/comm-central_ubuntu32_vm_test-xpcshell-bm02-tests1-linux32-build58.txt.gz
Summary: xpcshell-unpack.ini:toolkit/mozapps/extensions/test/xpcshell/test_signed_install.js (and signing failures) → TEST-UNEXPECTED-FAIL | xpcshell-unpack.ini:toolkit/mozapps/extensions/test/xpcshell/test_signed_install.js (and signing failures)
Might this be it?

https://treeherder.mozilla.org/#/jobs?repo=try-comm-central&revision=d193432e52b0
Assignee: nobody → mkmelin+mozilla
Comment on attachment 8609897 [details] [diff] [review]
bug1167986_signing_test_failures.patch

Review of attachment 8609897 [details] [diff] [review]:
-----------------------------------------------------------------

Seems try was happy
Attachment #8609897 - Flags: review?(standard8)
Attachment #8609897 - Flags: review?(standard8) → review?(Pidgeot18)
My understanding is that we were not going to require addon signing in Thunderbird. Doesn't this change enable required addon signing?
Comment on attachment 8609897 [details] [diff] [review]
bug1167986_signing_test_failures.patch

Review of attachment 8609897 [details] [diff] [review]:
-----------------------------------------------------------------

Yeah this is not what you want. You want to disable the test_signed_* tests when MOZ_ADDON_SIGNING isn't set. Not sure how to do that
Attachment #8609897 - Flags: review?(Pidgeot18) → review-
(In reply to Kent James (:rkent) from comment #10)
> My understanding is that we were not going to require addon signing in
> Thunderbird. Doesn't this change enable required addon signing?

Yes for official builds.
I don't think we ever discussed more than briefly. But I don't see why we wouldn't follow firefox's policy.
(In reply to Magnus Melin from comment #12)
> (In reply to Kent James (:rkent) from comment #10)
> > My understanding is that we were not going to require addon signing in
> > Thunderbird. Doesn't this change enable required addon signing?
> 
> Yes for official builds.
> I don't think we ever discussed more than briefly. But I don't see why we
> wouldn't follow firefox's policy.

We can discuss that, but the decision should not be buried in a bug that is purportedly about stopping a test failure.
Depends on: 1168571
Fair enough, filed bug 1168571. Frankly I didn't think that'd be a big thing - we change a lot of things all the times to align with firefox and avoid test failures. Given that it's a very contained patch I don't think there's much point in holding it up and having the tree all red while discussing. 

It's a pity all the config variables aren't available to do skips on for xpcxhell tests. If we want to disable the tests we probably just have to have them "skip-if = appname == "thunderbird"".
I don't understand the details, but can we disable this test or something?
This shows up in my local |make xpcshell-tests| and I am not sure if I can ignore them, and was about to submit a report when I found this bugzilla entry.

It seems that it boils down to enable/disable check on the signing of add-on.
Why not then we create one signed dummy add-on and non-signed dummy add-on and
see if
TB recognizes the signed add-on as such and
non-signed dummy add-on as such.
(Higher-level policy decision can be dealt with later?)
I mean, we simply check the low-level mechanism that supports such a policy decision.

TIA