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

Re: [D-I] Using Ubuntu's Rosetta or translate.org's Pootle for handling some Debian translations?

On Jan 20, 2005 at 08:31:55PM +0100, Christian Perrier wrote:

> The advantage for translators is obvious : the interface is very
> simple and involved people can focus on the real work : deal with PO
> files, and just forget about "nasty" stuff such as dealing with
> various Revision Control Systems or reporting bugs in various bug
> tracking systems.

That's my opinion too: especially for proof-readers a web-tool is a must.

> On the other hand, other people may have concerns about using an
> infrastructure controlled by a commercial company such as
> Canonical. The infrastructure itself is, as far as I know, not based
> on free software products (I may be wrong...this is not very clear to
> me).

Rosetta is definitely a closed project. At least for now. That makes it
seem a no-go to me. 

> This is not my own concern, but I feel it may be shared so
> I would not drive us into something which could make some people
> uncomfortable. Another argument is that the Pootle project seems to
> have been adopted (or on its way to be adopted) by several "key" FOSS
> projects and has already received a great interest in the i18n/l10n
> community.

I couldn't login into Pootle to give it a try, but I guess that it's not
an official translation platform for these apps, which rather makes is
even more difficult to use for efficient work. It's all the matter of
adoption of course and it's better than nothing.

> I just need your opinion about such ideas....as usually in Debian, the
> final decision will need to  be made on a consensus basis (or what
> will be closer to a consensus..:-))).
> We may also just decide to ignore these tools...or build our own (aka
> reinvent the wheel)....let's just discuss this.

I can't test Pootle just now, but I think it's about as (un)flexible as
Rosetta is. As it is GPL, is can surely be developed further to meet our

Some ideas for a web-based system:

- proper support for the complete gettext specification
- database storage
- keeping old strings in the database, just in case
- keep multiple translations of the same string in the DB, with
  explanation, why this particular translation is acceptable or not
- multiple access levels: Proof-reader, Translator, Coordinator. Only the
  Coordinators would have the right to "commit" the translations. All the
  other would go into the discussion queue.
- an external API. Useful for package maintainers, who submit their
  packages' translations to the system. update pot, fetch po etc. from the
  command line
- integration with CVS. Some more ideas needed, how this has to work
  considering permissions.
- string comparison: what has changed inside the string? There is a perl
  module that does this.
- xmlhttprequest-based interface ;)

Nikolai Prokoschenko 
nikolai@prokoschenko.de / Jabber: pronik@jabber.org

Reply to: