970478 - Certificate error pages are not completely translated in localized builds (missing pipnss.properties in build)

Status: RESOLVED INCOMPLETE

(Firefox for Android Graveyard :: General, defect, P5)

Description

[Alexander L. Slovesnik]

Attachment: Screenshot of bug

STR:
1) Install Fennec 28.0b1 from Google Play store
2) Switch language to Russian
3) Open page with Invalid certificate (e.g. https://seamonkey-project.org/)

Expected results:
Everything is translated

Actual results:
"Technical Details" section is in English, see screenshot

I've downloaded http://ftp.mozilla.org/pub/mozilla.org/mobile/releases/28.0b1/android/multi/fennec-28.0b1.multi.android-arm.apk and looked at it. Looks like it doesn't contain localized versions of http://mxr.mozilla.org/mozilla-central/source/security/manager/locales/en-US/chrome/pipnss/pipnss.properties
Perhaps, it's packaging issue.

Comment 1

[Francesco Lodolo [:flod]]

It's like this in nightly too for Italian. Should be another consequence of bug 792077.

Comment 2

[Alexander L. Slovesnik]

Hm, perhaps it's never been localized.

Comment 3

[Alexander L. Slovesnik]

Aaron, please care to read bug's intro and comments before moving it around. There is nothing to fix in Russian locale, everything is already translated - http://mxr.mozilla.org/l10n-mozilla-beta/source/ru/security/manager/chrome/pipnss/pipnss.properties#277 And Flod already have said that's it's a problem for Italian too.
It's bug in packaging, not in L10n.

Comment 4

[Aaron Train [:aaronmt]]

Let's re-summarize so it's clear ...

Comment 5

[Francesco Lodolo [:flod]]

Changing again subject: unless I'm wrong, issue is for all locales, not just ru and it (see bug 792077 and dependency for more info).

Comment 6

[Kevin Brosnan [Ex-Mozilla]]

Didn't we know this going out? 

The 28 Beta 1 build 1 was built with the Firefox 27 beta strings. After some investigation the primary UI did not show broken strings it was decided in release-drivers/channel-meetings that this was not critical, thus did not require a respin. See Axel's email to r-d "Firefox 28 Beta 1 built with state of 27, Desktop and Android"

Desktop avoided this due to some Windows 8 issues thus a build 2 was created.

Does http://ftp.mozilla.org/pub/mozilla.org/mobile/candidates/28.0b2-candidates/ resolve the issue? Builds have been kicked off and should be showing up over the next several hours.

Comment 7

[Alexander L. Slovesnik]

(In reply to Kevin Brosnan [:kbrosnan] from comment #6)
> http://ftp.mozilla.org/pub/mozilla.org/mobile/candidates/28.0b2-candidates/
> resolve the issue? Builds have been kicked off and should be showing up over
> the next several hours.
I've installed this build. It doesn't help.
Also it's problem in nightly, see comment #1

Comment 8

[Francesco Lodolo [:flod]]

Kevin, that's a different issue from the build problem in 28.0b1, it affects all branches.

In bug 792077 we stripped off all unnecessary strings, occasionally we realize we still need some of them (see dependencies of that bug).

Comment 9

[Axel Hecht [:Pike]]

Can someone help me out with the code paths here? I'd like to understand which strings of pipnss.properties are actually reachable here.

Comment 10

[Kevin Brosnan [Ex-Mozilla]]

Thanks for the explanation. We should get this fixed.

Comment 11

[Wesley Johnston (:wesj)]

(In reply to Axel Hecht [:Pike] from comment #9)
> Can someone help me out with the code paths here? I'd like to understand
> which strings of pipnss.properties are actually reachable here.

Looks like DocShell grabs these here:
http://mxr.mozilla.org/mozilla-central/source/docshell/base/nsDocShell.cpp#4491

which calls:
http://mxr.mozilla.org/mozilla-central/source/security/manager/ssl/src/NSSErrorsService.cpp#117

Comment 12

[Wesley Johnston (:wesj)]

Axel, can you take this? I see a few different sets of strings in:
http://mxr.mozilla.org/mozilla-central/source/security/manager/locales/en-US/chrome/

Not sure how many, which ones we need.

Comment 13

[Axel Hecht [:Pike]]

I don't really know either.

Brian, can you help?

My current thinking: in this bug, pack up all files referenced.

Get a security ui bug to actually clean up the UX. The strings that are together in the files are both strings that are human understandable (and that should be translated), and content only Brian understands. That should probably not be translated into Wolof or Fulah for people in Senegal, but be either in French or English, so that google can help. This is an effort we want to do across the desktop product, not just in this particular case.

Comment 14

[Sylvestre Ledru [:Sylvestre]]

Brian, could you have a look to this problem? Thanks

Comment 15

[Brian Smith (:briansmith, :bsmith, use NEEDINFO?)]

(In reply to Axel Hecht [:Pike] from comment #9)
> Can someone help me out with the code paths here? I'd like to understand
> which strings of pipnss.properties are actually reachable here.

Everything ahead of "# End of size restriction." should NOT be localized and everything after "# End of size restriction" SHOULD be localized.

Now, I think you may be asking "Which of these strings are used by Android and which ones are only used by Desktop Firefox?" I don't know the answer to that question and it would be a lot of work to find out a precise answer. I am also not sure why it matters? It seems like we should just default to using the desktop localizations of everything in Firefox for Android, and only create Android-specific localizations if/when we find a specific need to do so.

(In reply to Axel Hecht [:Pike] from comment #13)
> Get a security ui bug to actually clean up the UX. The strings that are
> together in the files are both strings that are human understandable (and
> that should be translated), and content only Brian understands. That should
> probably not be translated into Wolof or Fulah for people in Senegal, but be
> either in French or English, so that google can help. This is an effort we
> want to do across the desktop product, not just in this particular case.

Error messages from PSM usually/always include the error code, e.g. SEC_ERROR_REVOKED_CERTIFICATE. This is the Google-able part of the message. I am not an expert on translations but I would say that in general all of these natural-language strings should be translated fully or to the best judgement of the translators that know what people in the various locales want. If there are error messages that really "only Brian understands" then we should make sure that all those messages include an identifier like the SEC_ERROR_* or SSL_ERROR_* or whatever identifier.

I admit I don't quite understand what is being asked of me here but I hope the above helps.

Comment 16

[Sylvestre Ledru [:Sylvestre]]

Axel, are you happy with Brian's answers ? thanks
I am assigning the bug to you to make sure someone has this bug assigned.

Comment 17

[Axel Hecht [:Pike]]

(In reply to Sylvestre Ledru [:sylvestre] from comment #16)
> Axel, are you happy with Brian's answers ? thanks
> I am assigning the bug to you to make sure someone has this bug assigned.

Please don't assign bugs to me, given the cross-project nature of my job, I can't just take bugs from all parts of the organization. If I can arrange to work on something, I'll grab bugs. I'll look at Brian's reply later.

Comment 18

[Sylvestre Ledru [:Sylvestre]]

OK Axel. Sorry.

Chris, is there a chance that some files might not be included during the build process?

Comment 19

[Francesco Lodolo [:flod]]

(In reply to Sylvestre Ledru [:sylvestre] from comment #18)
> Chris, is there a chance that some files might not be included during the
> build process?

That's bug 792077.

Comment 20

[Axel Hecht [:Pike]]

See comment 8 for some of the reason.

To follow up on what Brian said:

We can't just include everything, as that's blowing our size constraints in the apk. Also bad for performance. The difference to desktop is not that we use different strings, but that we're shipping all languages in a single distributable bundle, while we only ship single-locale builds on desktop.

The stop-gap fix here is to find out which files are reachable here. Brian, is it really both pipnss.properties and nsserrors.properties? Then we'd need to do the same dance for security/ui that we do for the dom files in mobile/android/locales/jar.mn.

To get back to the other question: Think of Fulah. The elliptic curve strings will exist in Fulah exactly once, and that's as part of the Firefox localization. Thus searching for them on the web won't help. OTH, most-to-all Fulah speakers have some-to-good understanding of French, so using a French translation for those strings would be much better. The best way to communicate to the localizer which is which would be to have them in separate files, but that's the follow-up project.

Comment 21

[Lukas Blakk [:lsblakk] use ?needinfo]

This isn't really a release blocker and now we're out of time for FF28 so it will ship a second time.  Just nominate a fix for uplift when ready and it can go to the appropriate branches at that time based on risk v. reward.

Comment 22

[Brad Lassey [:blassey] (use needinfo?)]

filter on [mass-p5]

Comment 23

[Firefox Bug Husbandry Bot]

Re-triaging per https://bugzilla.mozilla.org/show_bug.cgi?id=1473195

Needinfo :susheel if you think this bug should be re-triaged.

Comment 24

[BMO Automation]

We have completed our launch of our new Firefox on Android. The development of the new versions use GitHub for issue tracking. If the bug report still reproduces in a current version of [Firefox on Android nightly](https://play.google.com/store/apps/details?id=org.mozilla.fenix) an issue can be reported at the [Fenix GitHub project](https://github.com/mozilla-mobile/fenix/). If you want to discuss your report please use [Mozilla's chat](https://wiki.mozilla.org/Matrix#Connect_to_Matrix) server https://chat.mozilla.org and join the [#fenix](https://chat.mozilla.org/#/room/#fenix:mozilla.org) channel.