Closed Bug 1237565 Opened 4 years ago Closed 4 years ago

TEST-UNEXPECTED-FAIL | C:\slave\test\build\tests\mozmill\composition\test-newmsg-compose-identity.js | test-newmsg-compose-identity.js::test_editing_identity

Categories

(Thunderbird :: General, defect)

45 Branch
defect
Not set

Tracking

(thunderbird45 fixed, thunderbird46 fixed)

RESOLVED FIXED
Thunderbird 46.0
Tracking Status
thunderbird45 --- fixed
thunderbird46 --- fixed

People

(Reporter: aleth, Assigned: aceman)

References

(Blocks 1 open bug)

Details

(Keywords: intermittent-failure, regression)

Attachments

(1 file)

05:04:54     INFO -  SUMMARY-UNEXPECTED-FAIL | test-newmsg-compose-identity.js | test-newmsg-compose-identity.js::test_editing_identity
 05:04:54     INFO -    EXCEPTION: Popup never opened! id=msgIdentityPopup, state=showing
 05:04:54     INFO -      at: utils.js line 447
 05:04:54     INFO -         TimeoutError utils.js:447 13
 05:04:54     INFO -         waitFor utils.js:485 11
 05:04:54     INFO -         _click_menus test-window-helpers.js:962 9
 05:04:54     INFO -         test_editing_identity test-newmsg-compose-identity.js:144 3
 05:04:54     INFO -         Runner.prototype.wrapper frame.js:585 9
 05:04:54     INFO -         Runner.prototype._runTestModule frame.js:655 9
 05:04:54     INFO -         Runner.prototype.runTestModule frame.js:701 3
 05:04:54     INFO -         Runner.prototype.runTestDirectory frame.js:525 7
 05:04:54     INFO -         runTestDirectory frame.js:707 3
 05:04:54     INFO -         Bridge.prototype._execFunction server.js:179 10
 05:04:54     INFO -         Bridge.prototype.execFunction server.js:183 16
 05:04:54     INFO -         Session.prototype.receive server.js:283 3
 05:04:54     INFO -         AsyncRead.prototype.onDataAvailable server.js:88 3
First occurred after https://hg.mozilla.org/comm-central/rev/d0e6db81879d. Please fix or back out.
Flags: needinfo?(mkmelin+mozilla)
Keywords: regression
Blocks: editablefrom
Oops, that patch was by aceman.
Flags: needinfo?(mkmelin+mozilla) → needinfo?(acelists)
Aceman is working on a fix. Sadly it's a very tricky one, the test works locally. I tried on Windows:

cd obj-i686-pc-mingw32/
mozmake SOLO_TEST=composition/test-newmsg-compose-identity.js mozmill-one

Result:

INFO Passed: 4
INFO Failed: 0
INFO Skipped: 0
SUMMARY-PASS | test-newmsg-compose-identity.js::setupModule
SUMMARY-PASS | test-newmsg-compose-identity.js::test_compose_from_composer
SUMMARY-PASS | test-newmsg-compose-identity.js::test_editing_identity
SUMMARY-PASS | test-newmsg-compose-identity.js::teardownModule

Note: I foolishly landed this on Aurora.
https://treeherder.mozilla.org/#/jobs?repo=comm-aurora&revision=a4907013150f

I will back it out there when I next land some stuff on Aurora.

BTW: I have a published query "comm-aurora checkin-needed":
https://bugzilla.mozilla.org/buglist.cgi?cmdtype=runnamed&namedcmd=comm-aurora%20checkin-needed&list_id=12778000
Yes, I am working on it, see my experiments at https://treeherder.mozilla.org/#/jobs?repo=try-comm-central&revision=5028354666d9 and older. However I am out of ideas for now. Nobody can reproduce it locally yet.

The pattern of clicking the menulist (using click_menus_in_sequence) is copied from other tests, which do not have the problem.

My theory was that the preceding SaveAs function is not yet finished when we try to click the identity menulist again so the click is eaten by the save progress bar. But I couldn't proof that yet.
Flags: needinfo?(acelists)
(In reply to :aceman from comment #4)
> My theory was that the preceding SaveAs function is not yet finished when we
> try to click the identity menulist again so the click is eaten by the save
> progress bar. But I couldn't proof that yet.

You could also dump compWin.e("msgIdentityPopup").state to see what the popup is doing before the call.

Maybe add a wait_for_window_focused(mainCompWin) as well?
(In reply to aleth [:aleth] from comment #5)
> (In reply to :aceman from comment #4)
> > My theory was that the preceding SaveAs function is not yet finished when we
> > try to click the identity menulist again so the click is eaten by the save
> > progress bar. But I couldn't proof that yet.
> 
> You could also dump compWin.e("msgIdentityPopup").state to see what the
> popup is doing before the call.

I could try that.
We know the state is "showing" AFTER the click. So that could indicate the click was not eaten, it went to the popup but that didn't initialize properly.
 
> Maybe add a wait_for_window_focused(mainCompWin) as well?

I tried that at https://hg.mozilla.org/try-comm-central/rev/b361e79e478f .
More experiments at:
https://hg.mozilla.org/try-comm-central/rev/3974b0c41941
https://hg.mozilla.org/try-comm-central/rev/e9ce597319f6
As discussed in bug 87987, let's disable the failing part of the test here. Then I make a new bug for making a working version.
Assignee: nobody → acelists
Status: NEW → ASSIGNED
OS: Unspecified → All
Hardware: Unspecified → All
Version: Trunk → 45
Blocks: 1238264
Attached patch patchSplinter Review
Disable the problematic part.
Attachment #8706020 - Flags: review?(rkent)
One question:

Do you want the test from bug 87987 and this fix in Aurora (TB 45) or do we just forget about the test for Aurora altogether?
(In reply to Jorg K (GMT+1) from comment #9)
> One question:
> 
> Do you want the test from bug 87987 and this fix in Aurora (TB 45) or do we
> just forget about the test for Aurora altogether?

We want to get tests working in Aurora, as they will usually propagate down to beta then esr. So no to "forget about the test for aurora?"
OK, so once this bug here is fixed, we can land it and the test from bug 87987 on Aurora. No problem.
Comment on attachment 8706020 [details] [diff] [review]
patch

Review of attachment 8706020 [details] [diff] [review]:
-----------------------------------------------------------------

This LGTM, and is test only. Tried the function locally and it seemed to work, so I am reasonably convinced this is probably a testing issue and not a real issue.

The parent bug was backed out of aurora, so that bug will need relanding before this is added as well.
Attachment #8706020 - Flags: review?(rkent) → review+
Comment on attachment 8706020 [details] [diff] [review]
patch

(In reply to Kent James (:rkent) from comment #12)
> The parent bug was backed out of aurora, so that bug will need relanding
> before this is added as well.
Sure, as soon as this is landed and has Aurora approval.
Attachment #8706020 - Flags: approval-comm-aurora?
Keywords: checkin-needed
https://hg.mozilla.org/comm-central/rev/eee6afc76d5afba2272f07c6151d6ac72cfdf89b
Bug 1237565 - disable failing part of mozmill\composition\test-newmsg-compose-identity.js::test_editing_identity. r=rkent
Status: ASSIGNED → RESOLVED
Closed: 4 years ago
Keywords: checkin-needed
Resolution: --- → FIXED
Target Milestone: --- → Thunderbird 46.0
Attachment #8706020 - Flags: approval-comm-aurora? → approval-comm-aurora+
You need to log in before you can comment on or make changes to this bug.