HEADS-UP! URGENT! Major problem with translations for Hardy and Intrepid.

[Bruno Patri]

Le Friday 16 January 2009 09:18:48 Milo Casagrande, vous avez écrit :

Hi,

> > Dear translation team leads,
> >
> > we have a major problem with translations in Hardy and Intrepid which
> > needs your attention NOW!
> >
> > It turns out that several translated msgstrings contain format strings
> > (i.e. %s, %d, etc.) where the msgid doesn't, or the number of format
> > attributes in msgstr and msgid does not match.
> >
> > This causes several applications to crash!
> >
> > We treat this as a severe regression and security issue which needs
> > urgent attention.
> >
> > We have prepared a list of potentially affected files (attached) and
> > need you to check all of those for this type of error and fix them
> > immediately.
> >
> > We cannot build new langpacks with those faulty strings present in the
> > database.
> >
> > Please see the attached list, investigate and fix the faulty strings.
> >
> > When finished, please report back which languages you have fixed.
> >
> > I cannot stress this enough: this is a severe regression in stable
> > releases. Please mobilize all resources to help fixing this problem.
>
> I understand and see the urgency in this case, but this is not going to
> be easy, really.
>
> Take for example, in my case, the "kdeprint" one... there's no package
> named that way in the "kdebase" super-package (and probably kde naming
> is not going to help us...).
>
> I supposed the number next to the package name was the faulty string
> number, probably it is, but not in Launchpad.
>
> Just one example:
>
> rosetta-hardy/it/LC_MESSAGES/inkscape.po.c-format:385:
>
> This is string 385 in Hardy Italian translation of Inkscape:
>
> http://tinyurl.com/9rc5g3
>
> Counting at 3465 strings, it's not going to be easy to search for an
> error which we have no idea of what is since that translation is 99.99%
> identical to upstream (only 3 strings have changed for convey to GNOME
> menu naming scheme and are not even c-format).
>
> Really, this is not going to be easy if it's proved to be true for all
> the packages listed.
>
> If that number is the string number, and such in my case there's no
> visible error, I really have no idea of what's going on.
>
> Is it possible to have a more detailed explanation of what those numbers
> mean or anything else that could cast some light on this issue?

This is the output of 

msgfmt -cv foo.po -o foo.mo

command. This is not the string number; this the line number in the .po file.

I've downloaded 2 files from rosetta (fr) : kcmlayout.po and kcmgreet.po. I 
didn't find any errors in the first one. There was an error at line 483 
although it's 562 in Arne's report....

It's nearly impossible to fix all this errors from Rosetta (or at least it's 
a huge work).
We need a direct acces to the .po files, download them, fix them and notfy 
upstream translators about those issues.



-- 
Bruno