Copying translations between releases

[David Planella]

Hi Jeroen,

here are my views on the proposal. First of all though, a couple of comments:

2008/1/29, Jeroen Vermeulen <jtv at canonical.com>:
[...]
>
> As you are probably aware, when we open a new Ubuntu release for
> translation, we start out with a copy of all the translations in the
> preceding release.  The preceding release remains the preferred
> translation target for some time, so it continues to improve after the
> new release is opened.
>

In theory it should improve, but it does not quite work like that. The
latest releases have had only one or two lang-pack updates (one for
Gutsy, IIRC), which means that after the last lang-pack update there
is no point in continuing working in the preceding release's
translations beyond the point on which the new release's ones are
open. After that they are neither released nor carried out to the new
release anymore.

My point is that unless there are lang-pack updates for it, working on
the preceding release once translations for the new one are open is
simply a waste of resources.

> In the future we would like to copy these improvements automatically: if
> a translation is updated in the older release but not in the newer
> release, the update will also happen in the newer release.  The
> advantages should be obvious.

In my oppinion it would be more useful if the changes would propagate
downwards, that is, corrections in translations in the current release
should change the same strings in previous releases.

This would be useful for getting those corrections or new translations
into previous LTS versions for example (assuming again that there are
lang-pack updates for the LTS release), since at least in the case of
our team, we only work on current releases -I still think it does not
make sense to translate a release other than the current one. To me,
it only makes sense in the case of critical translations errors, which
IMO can be anyway easily located and corrected on a case basis.

>
> There is one small risk: in principle it's possible that you want to
> translate the exact same string, in the same package, differently in two
> subsequent Ubuntu releases.  If that happens, and later someone comes
> and updates the translation in the older release, you will automatically
> get the new translation in the new release as well.
>

There is a risk, yes, and I do not especially like the fact that this
could happen. It adds more complexity to the translation process for
too little benefit. Translators want that the bulk of their work
consists of translations, not of file management.

> Would that be a problem to you?  Is the risk real, or imagined?  How
> does it compare to the benefits?

In my oppinion as a plain translator, I would prefer to see the
development efforts for Launchpad Translation focused on implementing
and fixing features that have been long wanted and are more important
instead of adding new ones we could live without.

Just to name a few of the long-known, long-wanted features I am referring to:

  * Translation search
  * Upstream coordination
  * Frequent language-pack updates

Regards,
David.