861356 - Fennec 21.0b2 multilocale build contains only en-US locale

Status: RESOLVED FIXED

(Release Engineering :: Release Requests, defect)

Description

[Alexander L. Slovesnik]

I've just updated to Fennec 21.0b2 through Google Play and my interface became English instead of Russian.
I've downloaded and unpacked http://ftp.mozilla.org/pub/mozilla.org/mobile/releases/21.0b2/android/multi/fennec-21.0b2.multi.android-arm.apk and looks like omni.ja archive contains only chrome/en-US folder. In comparison omni.ja from 21.0b1 contains a lot of folders with l10n.

Comment 1

[Axel Hecht [:Pike]]

Let's see what releng says.

Comment 2

[Aaron Train [:aaronmt]]

Looks like he's right.

http://hg.mozilla.org/build/buildbot-configs/rev/1873dc76f695

Comment 3

[Alex Keybl [:akeybl]]

(In reply to Aaron Train [:aaronmt] from comment #2)
> Looks like he's right.
> 
> http://hg.mozilla.org/build/buildbot-configs/rev/1873dc76f695

I've asked Tony to post-mortem with you all why checking at least one localization other than English isn't a part of beta sign-offs.

Comment 4

[Aaron Train [:aaronmt]]

Confirming users who update from any language can update just fine except they get en-US instead of the system language.

Comment 5

[Tony Chung [:tchung]]

(In reply to Alex Keybl [:akeybl] from comment #3)
> (In reply to Aaron Train [:aaronmt] from comment #2)
> > Looks like he's right.
> > 
> > http://hg.mozilla.org/build/buildbot-configs/rev/1873dc76f695
> 
> I've asked Tony to post-mortem with you all why checking at least one
> localization other than English isn't a part of beta sign-offs.

We're investigating from QA what happened.

Comment 6

[[e-mail] (:bhearsum)]

There was multiple levels of fail that caused this:
1) The l10n dashboard defaults to not giving us what we actually ship
2) Instructions weren't fully followed for getting the changesets from the dashboard
3) No system or human caught the error when the release was reviewed on ship it
4) The error wasn't caught in the resulting builds before we shipped

#3 can be addressed with some better verification in the RelEng automation. I filed bug 861842 on this.

In any case, the immediate issue is fixed.