1373807 - [mozmill] testWeeklyWithExceptionRecurrence.js failing on all platforms, Linux more than others

Status: RESOLVED FIXED

(Calendar :: Internal Components, defect, P2)

Description

[Jorg K (CEST = GMT+2)]

+++ This bug was initially created as a clone of Bug #1329957 +++

Comment 1

[Markus Adrario [:Taraman]]

running locally I saw, that in week view, where the failures occur the event boxes appear sequentially.
So most probably this is a timing issue again.

I'll prepare a patch that changes the logic to wait for the last event and then checks for the preceding ones.

Comment 2

[Markus Adrario [:Taraman]]

I have a patch ready, unfortunately tryserver does not seem to run the mozmill-tests on linux debug builds...

Comment 3

[Jorg K (CEST = GMT+2)]

Tom, can you please take a look:
https://treeherder.mozilla.org/#/jobs?repo=try-comm-central&revision=6f8003d16c46b22a8fbfe2bddef31653c6d6468c

No Mozmill run for Linux 32/64 debug. Maybe related to bug 1342828?

Comment 4

[Tom Prince [:tomprince]]

It looks like the try parser doesn't know about the chunked mozmill tests. (https://bugzilla.mozilla.org/attachment.cgi?id=8880478&action=diff)

Comment 5

[Jorg K (CEST = GMT+2)]

This test also fails on Windows debug:
https://treeherder.mozilla.org/#/jobs?repo=comm-central&revision=b244943f2b123b5ca13cdf913835a131cf6bd99c&selectedJob=109536084

Comment 6

[Jorg K (CEST = GMT+2)]

Can you do your testing on Windows debug since it fails there, too?

Comment 7

[Tom Prince [:tomprince]]

You can also change try request to be `-u mozmill,mozmill-1,mozmill-2` to get the mozmill tests to run on both linux and elsewhere. (Untested)

Comment 8

[Markus Adrario [:Taraman]]

In these try-runs, the respective test is not failing:
https://treeherder.mozilla.org/#/jobs?repo=try-comm-central&revision=9ff7bd753c8dc4dd33a45da767c28aeb1e2fedf2
https://treeherder.mozilla.org/#/jobs?repo=try-comm-central&revision=7d52df35197f288e51bb2aa60d2ad3e37333078d

Comment 9

[Markus Adrario [:Taraman]]

Attachment: Fix V1.0

This patch fixes the test as shown in the try-builds in the previous comment.

Comment 10

[Jorg K (CEST = GMT+2)]

Comment on attachment 8881946 [details] [diff] [review]
Fix V1.0

Review of attachment 8881946 [details] [diff] [review]:
-----------------------------------------------------------------

Sorry, I have no review rights in calendar/. If Philipp is too busy, perhaps redirect to someone else with review rights.

::: calendar/test/mozmill/cal-recurrence/testWeeklyWithExceptionRecurrence.js
@@ +177,2 @@
>      controller.assertNode(lookupEventBox("week", EVENT_BOX, null, 4, HOUR));
> +    controller.assertNodeNotExist(lookupEventBox("week", EVENT_BOX, null, 5, HOUR));    

Nit: Trailing space.

Comment 11

[Markus Adrario [:Taraman]]

Hrmpf. I linted it to prevent this. Obviously eslint is not working properly in my installation...

Comment 12

[Markus Adrario [:Taraman]]

Comment on attachment 8881946 [details] [diff] [review]
Fix V1.0

Shifting Review to makeyday, since Philipp seems to be busy right now.

Comment 13

[[:MakeMyDay]]

Comment on attachment 8881946 [details] [diff] [review]
Fix V1.0

Review of attachment 8881946 [details] [diff] [review]:
-----------------------------------------------------------------

Thanks, r+ with the whitespaces removed.

Comment 14

[Markus Adrario [:Taraman]]

Attachment: Fix V1.1

Patch with the fixed whitespace

Comment 15

[Markus Adrario [:Taraman]]

Pushed to c-c:
https://hg.mozilla.org/comm-central/rev/0170000376576d23e86bd8faa3e51c4e659f85c2

Comment 16

[Jorg K (CEST = GMT+2)]

Not a total success since the test failure is still there on Linux 32bit.
https://treeherder.mozilla.org/#/jobs?repo=comm-central&revision=0170000376576d23e86bd8faa3e51c4e659f85c2&selectedJob=114715679

It seems to be gone from Linux 64bit.

Comment 17

[Markus Adrario [:Taraman]]

The failure there hoever is an unhappy coincidence with another problem.

[1] shows
> EXCEPTION: Timeout waiting for modal dialog to open.
Which leads to a different issue. Sometimes the Event-Dialog fails to open in Time (>30s) on the debug-builds. Since this happens very rarely, I did not think about changing this timeout up to now.


[1]: https://treeherder.mozilla.org/logviewer.html#?job_id=114715679&repo=comm-central&lineNumber=14482