Closed Bug 932310 Opened 11 years ago Closed 11 years ago

Some UI strings in Firefox 24.1.0 ESR l10n builds are in English

Categories

(Release Engineering :: Release Requests, defect)

Product:
Release Engineering
Component:
Release Requests
Type:
defect
Priority:
Not set
Severity:
major

Tracking

(firefox-esr2425+ fixed)

Status:
RESOLVED WORKSFORME
Tracking Flags:
Tracking Status
firefox-esr24 25+ fixed

People

(Reporter: unghost, Unassigned)

References

Details

(Keywords: qaurgent)

Attachments

(2 files)

Screenshot of options windows with mixed en-US and ru strings
102.66 KB, image/png
Details
String difference between Firefox 24 and Firefox 17
252.43 KB, patch
Details | Diff | Splinter Review 
STR:
Download and install http://ftp.mozilla.org/pub/mozilla.org/firefox/releases/24.1.0esr/win32/fr/Firefox%20Setup%2024.1.0esr.exe or http://ftp.mozilla.org/pub/mozilla.org/firefox/releases/24.1.0esr/win32/ru/Firefox%20Setup%2024.1.0esr.exe

Expected results:
All UI strings are in French or in Russian

Actual results:
Some UI strings are in English, like File -> New Private Window, Help -> Firefox Health Report, entire section in Options - Advanced - Data Choices...
Does this happen only on these two locales or all are impacted? Can someone confirm 17esr's look fine and this happens on 24 alone ?
Keywords: qaurgent
I can confirm the builds mentioned above. Am currently looking at others.
To answer your second question, it appears that these strings are not present in 17esr and are new to 24esr. 

Still looking into the first question.
Also checked locales "it" and "es_ES" - same issue as reported locales.

So it might be safe to say all locales affected.
The core issue here is that the changeset used for l10n build generation for the esr24 builds was incorrect, hence the missing translation.
For the current release we have release noted the issue as a known issue : https://www.mozilla.org/en-US/firefox/24.1.0/releasenotes/ . Given the limited number of strings changes we are going to track this bug to resolve/verify the issue in a future release.
(In reply to bhavana bajaj [:bajaj] from comment #8)
> Given the limited number of strings changes we are going to 
> track this bug to resolve/verify the issue in a future release.

Could you elaborate what "future" means? Because we are getting complaints for half translated options in the Italian support forum, I guess other locales are in the same situation.
More strings than the few in the screenshot are affected: http://imgur.com/a/XDdFJ
Moreover, .xpi languages pack add-on (24.1.0esr) from ftp server ftp://ftp.mozilla.org/pub/mozilla.org/firefox/releases/24.1.0esr/win32/xpi/ are also affected.

Maybe it would be nice if these languages pack were updated? It will temporally fix this issue for 24.1.0esr?
(In reply to Francesco Lodolo [:flod] from comment #10)
> Could you elaborate what "future" means? Because we are getting complaints
> for half translated options in the Italian support forum, I guess other
> locales are in the same situation.

I don't know if we have metrics for ESR alone to tell us how many non en-US users we have, but if this doesn't qualify as a good reason to rebuild I don't know what does (complaints keep coming).
Flags: needinfo?(bbajaj)
Also many texts in German version are English.
Happens only in 24.1-ESR, not in 24.0-ESR and not in 25.
(In reply to bhavana bajaj [:bajaj] from comment #8)
> For the current release we have release noted the issue as a known issue :
> https://www.mozilla.org/en-US/firefox/24.1.0/releasenotes/ . Given the
> limited number of strings changes we are going to track this bug to
> resolve/verify the issue in a future release.

Is this really that hard to respin and do a chemspill for ESR?

Not having translated strings is a pretty major issue.

Can you please explain how this happened?
Hi, I was about to deploy Firefox ESR 24.0 fr (I didn't notice the .1 release) on 5000 PCs in my organization in the next few days. This would be a major issue for my users when receiving the .1 update. Adding myself to the CC list to know when repackaging.
(In reply to Mike Kaply (:mkaply) from comment #17)
> (In reply to bhavana bajaj [:bajaj] from comment #8)
> > For the current release we have release noted the issue as a known issue :
> > https://www.mozilla.org/en-US/firefox/24.1.0/releasenotes/ . Given the
> > limited number of strings changes we are going to track this bug to
> > resolve/verify the issue in a future release.
> 
> Is this really that hard to respin and do a chemspill for ESR?
> 
> Not having translated strings is a pretty major issue.

Given we do not auto-update esr users until 24.2.0 our analysis here is that the impact will be low keeping in mind a)that it's only the string changes that happened between FF17 and now b) since its a overlapping cycle basically pre-qualifying esr24. It's very rare that we chemspill without having either a massively impacting web-breakage or sec-critical issue.
Also, do note that all the expected localization will be in place for our upcoming esr
> 
> Can you please explain how this happened?
This was a manual operational error when submitting the l10n changsets used to generate the builds and we are identifying process improvements to put more safety checks to avoid this issue again.
Flags: needinfo?(bbajaj)
(In reply to Francesco Lodolo [:flod] from comment #10)
> (In reply to bhavana bajaj [:bajaj] from comment #8)
> > Given the limited number of strings changes we are going to 
> > track this bug to resolve/verify the issue in a future release.
> 
> Could you elaborate what "future" means? Because we are getting complaints
> for half translated options in the Italian support forum, I guess other
> locales are in the same situation.

Unless we have a security related issue leading to chemspill or identify any other major web breakage or gather and evaluate feedback that this issue has a major user impact which may lead to a dot release, 24.1.1 could be spun. Otherwise 24.2.0 is the next planned esr with a target release date of 12/09/2013 within next six weeks . 

Also I want to highlight that we overlap two versions of the esr's before auto-updating folks to the next esr(24.2) to help pre-qualify the builds and help find any major issues in advance.
(In reply to bhavana bajaj [:bajaj] from comment #20)
> Unless we have a security related issue leading to chemspill or identify any
> other major web breakage or gather and evaluate feedback that this issue has
> a major user impact which may lead to a dot release, 24.1.1 could be spun.

I'll ask again: do we have metrics about how many localized ESR users we have? Because this definitely looks like a "major user impact" to me.

> the impact will be low keeping in mind a)that it's only the string changes that happened between FF17 and now

I wouldn't say "only". If I didn't do something wrong, this a comparison of the difference between Firefox 24 and Firefox 17 for Italian, excluding all non Firefox related folders.
(In reply to Francesco Lodolo [:flod] from comment #21)
> Created attachment 825732 [details] [diff] [review]
> String difference between Firefox 24 and Firefox 17
> 
> (In reply to bhavana bajaj [:bajaj] from comment #20)
> > Unless we have a security related issue leading to chemspill or identify any
> > other major web breakage or gather and evaluate feedback that this issue has
> > a major user impact which may lead to a dot release, 24.1.1 could be spun.
> 
> I'll ask again: do we have metrics about how many localized ESR users we
> have? Because this definitely looks like a "major user impact" to me.
> 
> > the impact will be low keeping in mind a)that it's only the string changes that happened between FF17 and now
> 
> I wouldn't say "only". If I didn't do something wrong, this a comparison of
> the difference between Firefox 24 and Firefox 17 for Italian, excluding all
> non Firefox related folders.

Yes about half of 14% of the esr users might be impacted due to these obscure string missing translations. Also note the fact that this is a overlapping cycle without us having auto updated anybody to this version yet and so did not meet our bar for respin. Hence the justification to release note it as a known issue for the time being and wait and watch for feedback. We did discuss this issue in our channel meetings (which are open to all, so please feel free to join/represent) and had taken input across all teams including user advocacy. We took all the same metrics/analysis into account that we would normally put together for any of our products that are released and might be in a similar situation.
(In reply to bhavana bajaj [:bajaj] from comment #22)
> Yes about half of 14% of the esr users might be impacted due to these
> obscure string missing translations. 

Main options (sharing, updates, fonts, cookies, DNT, see http://imgur.com/a/XDdFJ), or strings in Sync don't exactly qualify as obscure strings.

"Half of 14% of the ESR users might be impacted": sorry but I don't get it. What does this mean? 
That 86% of ESR users are using en-US builds? Why just "half of 14%"?
(In reply to bhavana bajaj [:bajaj] from comment #22) 
> Hence the justification to release
> note it as a known issue for the time being and wait and watch for feedback.

How are you tracking feedback on this? I would expect most of the "feedback" to happen on local community forums and in result be flying under the radar of the group that qualifies builds for a respin.
> Also note the fact that this is a overlapping cycle without us having auto updated anybody to this version yet and so did not meet our bar for respin.

Are you saying users of the first Firefox 24 ESR didn't get updated to this ESR?


So if this had happened with regular Firefox (not the ESR), are you saying you wouldn't have done a chemspill to fix it?
(In reply to bhavana bajaj [:bajaj] from comment #22)
> Also note the fact that this is a overlapping cycle without 
> us having auto updated anybody to this version yet

If bugs during the overlap are less important, then we should always wait until the very last possible day, before switching from one ESR to the next. But this greatly increases the risk that issues are found after the overlap, rather than during the overlap, so the overlap is useless.

If this bug is irrelevant, why do you offer non-english versions at all?
(In reply to Zbigniew Braniecki [:gandalf] from comment #24)
> (In reply to bhavana bajaj [:bajaj] from comment #22) 
> > Hence the justification to release
> > note it as a known issue for the time being and wait and watch for feedback.
> 
> How are you tracking feedback on this? I would expect most of the "feedback"
> to happen on local community forums and in result be flying under the radar
> of the group that qualifies builds for a respin.

We're tracking the feedback in this bug and treating it as any other release issue. We decided not to take the time to respin 24.1 and requalify based upon the number of impacted users (very low) and the user impact (debatable). We instead decided to wait and see whether this was critical to the ESR community or blocking any deployments. This is normal process.

As of right now, we understand that some members of the l10n community disagree with Release Management's decision to make this a known issue. I can currently count the number of ESR user complaints on one hand though. If more ESR deployments express concern, we will strongly consider a 24.1.1.
(In reply to Alex Keybl [:akeybl] from comment #27)

> I can currently count the number of ESR user complaints on one hand though.

I can currently count the number of ESR user appreciation for not fixing this bug on no hand though.
We have one of those ESR deployments and this is quite a major concern for us... Not everybody will get this update though, so only a few hundred users are affected...
Alex - There are some extremely large ESR deployments (ours included) that have multi-language requirements for which this is big deal.
(In reply to Alex Keybl [:akeybl] from comment #27)
> (In reply to Zbigniew Braniecki [:gandalf] from comment #24)
> > (In reply to bhavana bajaj [:bajaj] from comment #22) 
> > > Hence the justification to release
> > > note it as a known issue for the time being and wait and watch for feedback.
> > 
> > How are you tracking feedback on this? I would expect most of the "feedback"
> > to happen on local community forums and in result be flying under the radar
> > of the group that qualifies builds for a respin.
> 
> We're tracking the feedback in this bug and treating it as any other release
> issue. We decided not to take the time to respin 24.1 and requalify based
> upon the number of impacted users (very low) and the user impact
> (debatable). We instead decided to wait and see whether this was critical to
> the ESR community or blocking any deployments. This is normal process.
> 
> As of right now, we understand that some members of the l10n community
> disagree with Release Management's decision to make this a known issue. I
> can currently count the number of ESR user complaints on one hand though. If
> more ESR deployments express concern, we will strongly consider a 24.1.1.


... so, here is my "ESR user complaint" for my french company which is about approximately 8000 users
Deployment is blocked because of this little bug.
So, please, if possible, make a new release
Thank you in advance.
This is not a little bug, this is a big bug.
Many mistakes and very visible.
I add.
-Error in contextual menu (tab bar).
-Error in urlbar (search or enter address).
-Error in exit window (do not ask next time).
And much more.
A new release is necessary and urgent!
As a ESR user, I also express my disappointment for this serious bug in the release of version 24.1.0 and even more for the decision not to correct it in short time.

I think this bug is exposing all the translators (that are volunteers and an important part of Mozilla community) of unjustified criticism, because many users will think it is their fault. Fixing this bug is also a way to show respect for their invaluable work.

Finally I fear that this bug will give an impression of lack of professionalism in environments where this quality is more required (the ESR versions are used for the most part in companies).
We have an unexpected ESR24 release coming up later this week - in it, we plan to resolve this l10n issue as well as a ride-along. Please stay tuned.
We tested Firefox 24.1.1 and found that there are strings that are still not localized. You can see what we tested and the results here: 
https://wiki.mozilla.org/Releases/Firefox_24_ESR/Test_Plan#Regression_Testing.
Not really sure how much of this is due to the system language, the not localized snippet sounds strange.

****
fr, th, fi, pl, it - text is in English on the Save dialog (ctrl+S), open dialog (ctrl+O), the Downloads pop-up (options - general - download files to...), print menu (ctrl + p), about:home - the messages under the search bar. 
****

What puzzles me is "not completetly localized Bookmarks menu" for locales like fr and es-ES (unless you're talking about actual bookmarks).
Thanks for the detailed report.

I think they're fine, actually, I've been diving into the details a bit.

I've cross-checked http://hg.mozilla.org/releases/l10n/mozilla-release/fr/rev/FIREFOX_24_1_1esr_RELEASE and friends, and they coincide with the 24 desktop non-esr release, as they should.

> fr, th, fi, pl, it - text is in English on the Save dialog (ctrl+S), open dialog (ctrl+O), the Downloads pop-up (options - general - download files to...), print menu (ctrl + p), 

These are native OS dialogs, and are thus not subject to the mozilla l10n process.

> about:home - the messages under the search bar.

These should be the promotional snippers, and are likely cached results in the profile. One had to actually test on a new profile for each locale to see the default snippet coming from the localization, and then idle for a while to see the server-side updates to that.

There's more details for more locales, much of which are either non-localized features like devtools in Finnish, or just missing strings. Thai for example wasn't up-to-date for 24, so that's getting a 23 localization status.

Then there's items like the help page on sumo, which isn't fully localized in some languages, but that's also unrelated to the problem we had in 24.1.0.

The bookmarks menu items are probably also profile-related, the default bookmarks are stored on profile creation in the language you use at the time. At least I couldn't find untranslated strings in the tagged revision for es-ES.

Really in-depth testing, thanks a lot.

I think we can resolve this bug now, but I'm not sure which resolution to actually choose. Alex?
(In reply to Francesco Lodolo [:flod] from comment #36)
> Not really sure how much of this is due to the system language, the not
> localized snippet sounds strange.
> 
> ****
> fr, th, fi, pl, it - text is in English on the Save dialog (ctrl+S), open
> dialog (ctrl+O), the Downloads pop-up (options - general - download files
> to...), print menu (ctrl + p), about:home - the messages under the search
> bar. 
> ****
> 
> What puzzles me is "not completetly localized Bookmarks menu" for locales
> like fr and es-ES (unless you're talking about actual bookmarks).

It's the "Most visited" item that is not localized here. Translating bookmarks in general wouldn't make sense. But should the default bookmarks be localized (About us, Getting involved etc.)?
They're part of the profile: if you create a profile with en-US and then switch to an Italian build, those elements will remain (like other bookmarks remain).

Try creating a new profile with the Italian build, they should be in Italian. If they're not, we have a problem ;-)
Marking this resolved WFM per comment #37 and is verified on 24.1.1 esr.
Status: NEW → RESOLVED
Closed: 11 years ago
status-firefox-esr24: affected → fixed
Resolution: --- → WORKSFORME
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: