l10n approach for bzr

[James Westby]

On Mon, 2008-03-17 at 15:56 -0500, Mark Hammond wrote:
> Unless I'm missing something, it is already worse than that - you need to
> locate *all* existing translations and add the new insert (else a runtime
> error formatting the string will occur in some locales but not others) - and
> worse, you need to know *where* in the translated text the insert makes
> sense?

That's a very good point.

> From a practical POV, when someone makes a change to the English version of
> a string, should we ask them to locate that string in *all* translations and
> remove them, to make it clear the translation is no longer used?  I doubt we
> would want the translation files to end up with the complete set of strings
> ever used now or in the past with no indication if the string is actually
> used in the current version.  Or is there another alternative that can help
> us keep these things in synch over time and detect old, unused translations
> for each locale?

That shouldn't be an issue. It's a well known problem in l10n. If I 
change the msgid then all the translations a marked as "fuzzy" so that
the translator knows that they need to fix them up.

I think fuzzy strings are also not shown to the user. This is obviously
not great in some cases, but at least it shields them from string
formatting exceptions caused by a change.

Thanks,

James