1026582 - Test failure in testSessionStore.testUndoTab.testUndoTabViaShortcut for [hsb] locale

Status: RESOLVED FIXED

(Mozilla QA Graveyard :: Mozmill Tests, defect)

Description

[Liz Henry (:lizzard Please n-i to RyanVM, jcristau, or pascal)]

Functional tests for the hsb locale for Firefox 31.0b2 are failing for the testSessionStore.testUndoTab.testUndoTabViaShortcut tests.

http://mozmill-release.blargon7.com/#/functional/failure?app=Firefox&branch=31.0&platform=Win&from=2014-06-10&to=2014-06-17&test=%2FtestSessionStore%2FtestUndoTab.js&func=testUndoTabViaShortcut

Comment 1

[Francesco Lodolo [:flod]]

I'm looking at the code but I'm not sure I understand how the localization could create the failure. Aren't you using the IDs of the context menu to activate commands?

Comment 2

[Francesco Lodolo [:flod]]

Out of curiosity I just downloaded the build linked in https://bugzilla.mozilla.org/show_bug.cgi?id=1018198#c14 for OS X: both menu and shortcut work fine to reopen the last closed tab.

Comment 3

[Henrik Skupin (:whimboo) [⌚️UTC+1]]

(In reply to Francesco Lodolo [:flod] from comment #2)
> Out of curiosity I just downloaded the build linked in
> https://bugzilla.mozilla.org/show_bug.cgi?id=1018198#c14 for OS X: both menu
> and shortcut work fine to reopen the last closed tab.

The problem we haven't seen on OS X, but only for Windows. Please see the report URL as given by Liz in comment 0. The DTD entities look fine, so maybe we are facing an issue (like bug 1005942 for Firefox 29.0) in reopening recently closed tabs.

Andreea, it would be good if someone could check that.

Comment 4

[Cosmin Malutan, [:cosmin-malutan]]

This fails because we have an conflict in access keys in browser.dtd:
tabCmd.commandkey "t" === pageSourceCmd.accesskey "t"
http://mxr.mozilla.org/l10n-mozilla-beta/source/hsb/browser/chrome/browser/browser.dtd#82

Instead of restoring a the previous closed tab here it opens the page source, this can be observed visually when running the test.

Comment 5

[Francesco Lodolo [:flod]]

I'm lost in Mozmill territory. From the first link, I get here
http://mozmill-release.blargon7.com/#/functional/report/6959f9a8236961096763389083a752b8

Which tells me that the failing test is "testUndoTabViaShortcut".

The only method that seems interesting is tabBrowser.reopen("shortcut")
http://hg.mozilla.org/qa/mozmill-tests/file/7dc8bfc3d8014c242f611f6501a7d16c245ffca4/firefox/tests/functional/testSessionStore/testUndoTab.js#l99

But the JS seems to use tabCmd.commandkey
http://hg.mozilla.org/qa/mozmill-tests/file/db8b8621c485/firefox/lib/tabs.js#l910

I don't see any relation between a commandkey and an accesskey.

Comment 6

[Henrik Skupin (:whimboo) [⌚️UTC+1]]

Me neither. Not sure why this assumption has been made. Both are totally independent, and wont harm each other.

Comment 7

[Cosmin Malutan, [:cosmin-malutan]]

It actually toggles the webConsole, and it should restore an closed tab, I assumed it was a conflict.  
http://mxr.mozilla.org/l10n-mozilla-beta/source/hsb/browser/chrome/browser/browser.dtd#267

Comment 8

[Michael Wolf]

In bug 1026990 - https://bugzilla.mozilla.org/show_bug.cgi?id=1026990 - I wrote yesterday that I thought that the keytext key for the web console is an accesskey and changed it. I translate on Pootle an changed the key for Fx 32.0a2 Aurora there. But I don't have access to mozilla-central so I ask that someone of you changes this for Fx 31.0b2. Thanks

Comment 9

[Henrik Skupin (:whimboo) [⌚️UTC+1]]

Michael, we have to keep that conversation on bug 1026990. Over there I requested help from Francesco. So lets see that we can get this in soon. Thanks for your fix!

Comment 10

[Michael Wolf]

Hello Henrik, I saw this bug by chance and since I didn't got a reply on bug 1026990, I wanted to be on the safe side, true to the motto: Two are better than one. :-)

Comment 11

[u279076]

Failure is still happening in the latest Beta builds.

Comment 12

[Liz Henry (:lizzard Please n-i to RyanVM, jcristau, or pascal)]

This failure is still happening for functional tests for Beta 31 RC: http://mozmill-release.blargon7.com/#/functional/report/a3b8d1e9bea5a0fc05cd41dbf63c4945

Comment 13

[Michael Wolf]

Hello Liz, hello Anthony. Who can change it? I have right for Pootle only but Pootle translations are for Fx 32 Aurora. See bug 1026990:

https://bugzilla.mozilla.org/show_bug.cgi?id=1026990#c7
https://bugzilla.mozilla.org/show_bug.cgi?id=1026990#c9

Or can I get the rights for mozilla-central to change it myself? I don't want that the Upper Sorbian Firefox 31 is not released because of this bug.

Comment 14

[Francesco Lodolo [:flod]]

Michael, don't worry about mozilla-central: hsb/dsb are not even available there.

Also, it's too late to accept new changes for mozilla-beta, and I think hsb it's already included in the release that will come out next week. If you fixed it on aurora, it will move to beta next week automatically.

Comment 15

[Michael Wolf]

Thank you Francesco. I am glad to hear that. But, I changed it for Fx 32 Aurora only because the bug was filed only after Fx 32 has already been on Pootle.

Comment 16

[Liz Henry (:lizzard Please n-i to RyanVM, jcristau, or pascal)]

This failed again today for Firefox 31.0 hsb on mm-win-81-32-3 (2014-07-17_12-39-55). The failed test is testUndoTabViaShortcut from /testSessionStore/testUndoTab.js with the error message "New tab has been opened". 

http://mozmill-release.blargon7.com/#/functional/report/db75145408bff5ccf8a072085b4d897b

Comment 17

[Michael Wolf]

Who can change that still for Fx 31? Who creates the builds? What is the used source?

Comment 18

[Henrik Skupin (:whimboo) [⌚️UTC+1]]

Given Francesco this is a wontfix for Firefox 31, given that we are too late in the cycle. But might be able to fix the bug in the next esr release? Francesco, is that possible?

Comment 19

[Michael Wolf]

A question: Is it possible to fix the bug at least for the Fx 31 language pack?

Comment 20

[Henrik Skupin (:whimboo) [⌚️UTC+1]]

(In reply to Michael Wolf from comment #19)
> A question: Is it possible to fix the bug at least for the Fx 31 language
> pack?

Lets ask Axel and Flod, who should know.

Comment 21

[Francesco Lodolo [:flod]]

As I said the sign-off process ended last Monday for Firefox 31, that includes both the build and the language pack (there are no separate sign-offs).

As far as I know, the same thing is valid for Firefox ESR, but I'm not completely familiar with the process.

Comment 22

[Michael Wolf]

Yes, but the bug has already been filed on June 17, 2014 and already on June 19, 2014, 2 days later, I've given the solution. Since then nothing happened, except that someone has pointed out that the bug is not fixed yet.

Comment 23

[Axel Hecht [:Pike]]

We're not going to get this fixed in 31, and we're not taking updates for localizations on ESR either.

The fix for this bug is going to be in 32, I verified that by looking at the source we'll ship to betas next week.

Resolving this bug as FIXED, updated the flags.