[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Debian Debconf Translation proposal ( again )

On Wed, Jan 07, 2004 at 05:47:53PM +0100, Dominique Devriese wrote:
> Colin Watson writes:

> >> I just wanted to say it would be a little bit easier, if
> >> translation wouldn't have to pass through a package maintainer (for
> >> both translator/maintainer).

> > And it would be less good; maintainers do pick up errors in
> > translations from time to time, which is valuable. 

> FWIW, I've never seen that happen.

It happens.  In practice, I proofread about 85% of the debconf
translations that are submitted on my packages, and have found typos and
other errors in a handful of them.  Even on languages I can't read, such
as Japanese, there have been cases where a consistency check between two
almost-identical templates in two related packages has shown the
translators that additional coordination and improvement was possible.
While I'm probably an extreme case, I believe the majority of our
developers are bilingual (at least), which means most maintainers are
capable of reviewing at least *some* of the translations submitted on
their packages.  I think we all agree that review is good, and
maintainers are among the *best* people to have reviewing for you if
they know the language, because they have a personal interest *in this
particular translation* that few other reviewers will have.

> However, errors in translations will be picked up in the new
> situations too, by the people actually using the translations.

In a few of the corrections I requested of the translator, the
translation was technically very accurate and would not be noticed as
wrong by a user, but lost a nuance of the original text that I felt was
important to communicate.

> > One of the jobs of a maintainer is to collate contributions to a
> > package. 

> Here you implicitly define translations to be part of a package.
> Another definition might put translations in one single different
> package.

Today, Debian is principally divided into task- or program-oriented
units.  This means that the maintainer is responsible for providing a
*cohesive functional unit* containing documentation, source code,
package scripts, and many other things that he or she may not be
personally competent to make changes to or review for correctness.
Including translations is just one more part of a maintainer's

If you have read the thread you've been directed to about the DDTP, you
should already be familiar with these arguments.

> > Please stop trying to subvert this or work around it! 

> I don't think the little amount of extra checks on the translations is
> worth the enormous administrative workload on the translators.  Please
> reread Tim Dijkstra's post, and imagine having to do that checklist
> 10000 times for every single Debian package. ( Not sure about the
> number ).

To date, all proposals for moving debconf translations out of packages
have included in their rationale that maintainers will no longer be
responsible for merging translations.  I don't think I've seen a single
maintainer speak up so far to agree that he wants this burden removed
from him; only translators seem to think this is important, and only
because of maintainers who are not responsive to bug reports.  We should
not be seeking to absolve non-responsive maintainers of their duty to
merge translations, we should be identifying such maintainers and
addressing the maintenance problems.  The only workload described in
Tim's checklist that would go away under a different paradigm is that of
NMUing packages, and it doesn't go away just because the translators
aren't the ones prompting the NMUing.

Steve Langasek
postmodern programmer

Attachment: signature.asc
Description: Digital signature

Reply to: