1137575 - 10.10|10.14 bustage in test_native_mouse_mac.xul

Status: RESOLVED FIXED

(Core :: Widget: Cocoa, defect)

Description

[Phil Ringnalda (:philor)]

test_native_mouse_mac.xul was doing fine when we were staging running tests on 10.10.0, but then at what in retrospect was too much of the last minute, we switched to 10.10.2, and now we have very frequent failures a la

https://treeherder.mozilla.org/logviewer.html#?job_id=5224775&repo=try

23:44:26 INFO - 2771 INFO TEST-UNEXPECTED-FAIL | widget/tests/test_native_mouse_mac.xul | received event I didn't expect: {"type":"mouseout","target":"box","screenX":150,"screenY":150} - expected PASS
23:44:26 INFO - 2772 INFO TEST-UNEXPECTED-FAIL | widget/tests/test_native_mouse_mac.xul | Didn't receive expected event: {"type":"mousemove","target":"box","screenX":170,"screenY":150} - expected PASS
23:44:26 INFO - 2773 INFO TEST-UNEXPECTED-FAIL | widget/tests/test_native_mouse_mac.xul | received event I didn't expect: {"type":"mouseover","target":"box","screenX":170,"screenY":150} - expected PASS
23:44:26 INFO - 2774 INFO TEST-UNEXPECTED-FAIL | widget/tests/test_native_mouse_mac.xul | received event I didn't expect: {"type":"mousemove","target":"box","screenX":170,"screenY":150} - expected PASS

and we're stuck too far along in the transition from 10.8 to 10.10 to be able to turn back, so I need to temporarily disable it while we figure out whether it really was a 10.10.0->10.10.2 change that broke us, or some sort of infra bustage, or what. Lucky for me, Markus was a bit surprised we were still running it at all after its early history of failure, rather than being horrified at losing it.

Comment 1

[Kim Moir [:kmoir] ET]

Attachment: bug137575.patch

Comment 2

[Kim Moir [:kmoir] ET]

:mstange I can't seem to land on inbound (permission denied) Could you land this for me?

Comment 3

[Kim Moir [:kmoir] ET]

Actually I'll get the sheriffs to land it, that would be easier

Comment 4

[Ryan VanderMeulen [:RyanVM]]

Pushed to inbound per Kim's IRC ping.
https://hg.mozilla.org/integration/mozilla-inbound/rev/e3138fd47c7d

Comment 5

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/e3138fd47c7d

Comment 6

[Kim Moir [:kmoir] ET]

How do I get this uploaded to aurora?  I ran the tests on my dev-master today and the tests still failed on aurora.  The goal for Q1 is to get 10.10 tests running on trunk and aurora.

Comment 7

[Ryan VanderMeulen [:RyanVM]]

I've added it to my queue for the next time I'm pushing to Aurora. Probably won't be today, though.

Comment 8

[Kim Moir [:kmoir] ET]

thanks!

Comment 9

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/9d2a26109463

Comment 10

[Kim Moir [:kmoir] ET]

I think this can be closed.

Comment 11

[Phil Ringnalda (:philor)]

Not really. The test is still disabled, annotated with this bug being the explanation of why, and now 10.10 is the only version we run on so the test is just completely disabled. That leaves the options of having this bug open, or removing the test from the tree.

Comment 12

[Phil Ringnalda (:philor)]

Current state of the world is https://treeherder.mozilla.org/#/jobs?repo=try&revision=7ba1740d50686a4a41bb65b7695eb72acbfb895a&group_state=expanded

We consistently get an NS_ERROR_FAILURE from nsIDOMWindowUtils.sendNativeMouseEvent the first time we try to touch the inactive window on the right side, in https://hg.mozilla.org/mozilla-central/annotate/8b1f0ca44c7be9a15bfbedf0cde540831f2e2aec/widget/tests/native_mouse_mac_window.xul#l262

On opt, we then hit https://hg.mozilla.org/mozilla-central/annotate/8b1f0ca44c7be9a15bfbedf0cde540831f2e2aec/widget/tests/native_mouse_mac_window.xul#l176 and bail, on debug we sit until we time out.

Screenshots on opt consistently do not show the right window, and either do show the left window or also don't show it (thus leaving the mouse maybe just below or maybe on the parent window's "Run Chrome Tests" button, a slightly scary place to be). Screenshots on debug will maybe show both windows, or neither with the mouse by the button, or neither with the mouse further south and east and looking like it thinks it is over a link it isn't over. No idea whether they are showing anything real, though, since that could easily mean they are taken partway through onTestsFinished() on opt, and at a random spot in the harness cleaning up on debug.

Comment 13

[Joel Maher ( :jmaher ) (UTC -8)]

this is broken as well in osx 10.14 (as referenced in bug 1550112)

Comment 14

[Joel Maher ( :jmaher ) (UTC -8)]

Attachment: Bug 1137575 - disable test_native_mouse_mac.xul for osx 10.14.

disable test_native_mouse_mac.xul for osx 10.14.

Comment 15

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/0ac1469f9416
disable test_native_mouse_mac.xul for osx 10.14. r=spohl

Comment 16

[Andrei Ciure[:aciure]]

https://hg.mozilla.org/mozilla-central/rev/0ac1469f9416

Comment 17

[BugBot [:suhaib / :marco/ :calixte]]

The leave-open keyword is there and there is no activity for 6 months.
:spohl, maybe it's time to close this bug?

Comment 18

[Stephen A Pohl [:spohl]]

(In reply to Release mgmt bot [:sylvestre / :calixte / :marco for bugbug] from comment #17)

The leave-open keyword is there and there is no activity for 6 months.
:spohl, maybe it's time to close this bug?

Comment 11 still applies.

Comment 19

[BugBot [:suhaib / :marco/ :calixte]]

The leave-open keyword is there and there is no activity for 6 months.
:spohl, maybe it's time to close this bug?

Comment 20

[Stephen A Pohl [:spohl]]

(In reply to Release mgmt bot [:sylvestre / :calixte / :marco for bugbug] from comment #19)

The leave-open keyword is there and there is no activity for 6 months.
:spohl, maybe it's time to close this bug?

Comment 11 still applies.

Comment 21

[BugBot [:suhaib / :marco/ :calixte]]

The leave-open keyword is there and there is no activity for 6 months.
:spohl, maybe it's time to close this bug?

Comment 22

[Stephen A Pohl [:spohl]]

Markus, do you remember these tests and would you know what needs to be done here to reenable them?

Comment 23

[Markus Stange [:mstange]]

I remember them, but testing widget behavior so close to the OS is always brittle. I also haven't seen any regressions in this code in the last eight years, so the test is probably not paying for itself. We could just remove it.

Comment 24

[Stephen A Pohl [:spohl]]

Attachment: Bug 1137575: Remove macOS native mouse event tests due to brittleness. r=mstange

Comment 25

[Stephen A Pohl [:spohl]]

Thank you for your input, Markus!

Comment 26

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/567347cd1c51
Remove macOS native mouse event tests due to brittleness. r=mstange

Comment 27

[Atila Butkovits]

https://hg.mozilla.org/mozilla-central/rev/567347cd1c51

Comment 28

[Atila Butkovits]

https://hg.mozilla.org/mozilla-central/rev/567347cd1c51