669835 - Test failure 'could not validate element Name: q with value area code' in testGoogleSuggestions.js

Status: RESOLVED FIXED

(Mozilla QA Graveyard :: Mozmill Tests, defect)

Description

[Henrik Skupin [:whimboo][⌚️UTC+2]]

Module:    testGoogleSuggestions.js
Test:      testGoogleSuggestedTerms    
Failure:   could not validate element Name: q with value area code
Branches:  ALL
Platforms: ALL

This regression has been started on June 15th and happens across all platforms. Could be a change made by Google.

Aaron, would you be able to check that failure? If not we should ask Vlad or Alex.

Comment 1

[Henrik Skupin [:whimboo][⌚️UTC+2]]

Aaron, this is in one of your owned areas and needs investigation and a fix. Can you please take care of it?

Comment 2

[Aaron Train [:aaronmt]]

This test passes for me on all my local platforms, I'll check into this.

Comment 3

[Henrik Skupin [:whimboo][⌚️UTC+2]]

if you can't reproduce just let it run in a loop with over 100 or so iterations. Just to prove the tests stability.

Comment 4

[[:Aleksej]]

I have a Russian IP, and http://google.com redirects to Russian page http://google.ru with an "en-US" or "ru" build. With an "eo" build, it just shows Esperanto page at http://google.com, so the test doesn’t fail.

Comment 5

[[:Aleksej]]

(In reply to comment #4)
As Aaron pointed out, that’s probably not relevant, because the test uses https://encrypted.google.com and changes "intl.accept_languages" to "en-us".

Comment 6

[Aaron Train [:aaronmt]]

I can't reproduce this over 120 iterations across Ubuntu 11.04, and Windows 7, not sure where to take this.

Comment 7

[Henrik Skupin [:whimboo][⌚️UTC+2]]

(In reply to comment #6)
> I can't reproduce this over 120 iterations across Ubuntu 11.04, and Windows
> 7, not sure where to take this.

On qa-horus? If not you should test on this machine because from there it always fails as the todays reports show again.

Comment 8

[Henrik Skupin [:whimboo][⌚️UTC+2]]

Attachment: Patch v1

nodeCollector is not the best choice for this test due to the dynamic content. It really needs to be updated soon. Now with the Selector method available on the elementslib lets use that one instead. It fixes the problem for me on Vista.

Comment 9

[Geo Mealer [:geo] -- This account is inactive after 2015-07-07]

Comment on attachment 547238 [details] [diff] [review]
Patch v1

Looks good, r+. 

Reminds me that I still need to add a waitFor-type function onto the refactored elements too, for times when you want something more than implicit wait.

Comment 10

[Henrik Skupin [:whimboo][⌚️UTC+2]]

This test drives me crazy. The Google website doesn't behave reliable enough that we really can test this with Mozmill. I have massive problems with the backport to older branches. When the test fails once out of a dozen of testruns. I have absolutely no idea how to fix it. Even a call to tabs.activeTab.querySelector doesn't return the suggestion after a long delay.

Geo, I would propose that we disable this test.

Comment 11

[Henrik Skupin [:whimboo][⌚️UTC+2]]

I should CC Geo if I want to get feedback from him.

Comment 12

[Geo Mealer [:geo] -- This account is inactive after 2015-07-07]

Heh, thanks. Yeah, easier for me to see it that way. ;)

If it's breaking constantly, we should pull it. I don't think this test is high enough value to merit a lot of time and trouble.

Comment 13

[Henrik Skupin [:whimboo][⌚️UTC+2]]

Ok, I have removed the test from all branches:
http://hg.mozilla.org/qa/mozmill-tests/rev/484f33c19816 (default)
http://hg.mozilla.org/qa/mozmill-tests/rev/56a42fea0daa (aurora)
http://hg.mozilla.org/qa/mozmill-tests/rev/a99fbf999d20 (beta)
http://hg.mozilla.org/qa/mozmill-tests/rev/2ca0bd80d6a5 (release)

Litmus tests have been updated accordingly.

Comment 14

[Henrik Skupin [:whimboo][⌚️UTC+2]]

Sorry, but I missed to remove the patch from 2.0 and 1.9.2. Did that now:
http://hg.mozilla.org/qa/mozmill-tests/rev/bb642d1265b8 (2.0)
http://hg.mozilla.org/qa/mozmill-tests/rev/bb97f7eed669 (1.9.2)