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

Re: perl / gettext and http://www.iro.umontreal.ca/contrib/po/Po/

Mistyped the address, so this email didn't get through before.

- --DAA03717.935048383/amsterdam.interport.net
Content-Type: message/rfc822

Return-Path: <apharris@burrito.onshore.com>
Received: from burrito (mail@ts1port9.port.net [])
	by amsterdam.interport.net (8.8.5/8.8.5) with ESMTP id DAA03712;
	Thu, 19 Aug 1999 03:39:38 -0400 (EDT)
Received: from apharris by burrito with local (Exim 2.11 #1 (Debian))
	id 11HMnE-0000G9-00; Thu, 19 Aug 1999 03:39:24 -0400
To: Fran ois Pinard <pinard@iro.umontreal.ca>
cc: debian-boot@lists.deian.org, debian-doc@lists.debian.org
Subject: Re: perl / gettext and http://www.iro.umontreal.ca/contrib/po/Po/ 
From: Adam Di Carlo <adam@onshore.com>
In-reply-to: <oqn1vobebt.fsf@titan.progiciels-bpi.ca> 
References: <oan1vooc52.fsf@burrito.fake> <oqn1vobebt.fsf@titan.progiciels-bpi.ca>
Comments: In-reply-to Fran ois Pinard <pinard@iro.umontreal.ca>
 message dated "19 Aug 1999 00:10:30 -0400."
Date: Thu, 19 Aug 1999 03:39:24 -0400
Message-Id: <E11HMnE-0000G9-00@burrito>
Sender: Adam Di Carlo <apharris@burrito.onshore.com>
MIME-Version: 1.0

>Adam Di Carlo <adam@onshore.com> writes:
>> I maintain the Debian Installation Manual, and I need to translate gettext
>> (PO) files into SGML entities.
>I'm not fully sure of your goal.  For what I guess I understand (:-),
>it looks like a simple job, and if it might be useful for me as well,
>we can work it together (you the specs, me the code, roughly, maybe).
>If you are not afraid of Perl, I guess it would be simple and comfortable
>for you to do it on your side without having to wait after me.

Actually, what I'm trying to do is probably simple and definately
general.  We have some programs which I am documenting; these programs
use gettext for the messages they present to the user.  My intention
is to use the gettext catalog itself in the SGML documentation of the
program.  This is a big win especially since the gettext messages, and
the manuals are already translated into about 12 languages ( Croatian,
Czech, English, Finnish, French, German, Japanese, Korean, Spanish,
Portuguese and Russian), with more to come.

So, I need to pick up the right gettext catalog (PO) for the language
I am building the documentation for, and convert those into some sort
of data I can reference from SGML.  It seems to me the easiest way
would be to take gettext message id's (skipping msgfmt -- is that
wise?) and turn those into SGML entity names (I'd have to munge them a
bit so they fit into the legal set of SGML entity name characters).

>> I've looked over several Perl packages, but honestly, your stuff from
>> http://www.iro.umontreal.ca/contrib/po/Po/ looks like the best bet for
>> doing it.  Do you have suggestions?
>I would use one or two modules from po/Po/, without changing them.  You may
>find a few simple examples of this in bin/.
>> How is that code licensed?  Is it distributable?  Under the GPL?
>I do not remember having assigned these to the FSF, I nevertheless hope
>this is free enough for you.  Consider it distributable under the GPL.
>I would not mind even free-er conditions.  I may add a few legalistic
>lines in the sources, wherever you think you need them.

I don't ask that you assign them to the FSF, I only ask they they be
clearly marked with some sort of copyright that allows me to use,
distribute, and modify them.  I.e., you could put this on the top of
your files:

# This program is free software under the terms of the GNU General Public
# License (GPL). A copy of the GPL, "COPYING", should have been made
# available with this software.  If not, a copy may be obtained at 
# http://www.fsf.org/copyleft/gpl.html

>My only (informal) demand is that you try to respect the Translation
>Project.  I know that Debian often tries to duplicate responsibilities on
>its side, creating some confusion at times.

Well, it *has* happened, but I don't think the project
self-consciously sets out to do this.  I for one have always lobbied
hard for more coordination, especially with external documentation
efforts (it's such a big big job!).

The project in particular I am asking this on behalf of is the Debian
GNU/Linux Installation Manual, browsable at

>I hope Debian will not try to short-circuit the TP.  Not at all that
>I'm starving for control.  This is just that if some national teams
>are strong, others are a bit weak and would not welcome confusion.
>My only goal is that we all move more solidly towards
>internationalised/localised software; collaboration does it better.

This is our goal as well.  Also, even though we don't always clearly
express it, it is our goal to collaborate and contribute *upstream*
and not just "fork" material off on the Debian package level.

My primary concern is pure Debian documentation (I work mostly in
SGML) and the like, however, I am interested to see that the TP is
happy with Debian, and that we can all fruitfully contribute.

Reading some of the material from your site
<http://www.iro.umontreal.ca/~pinard/po/HTML/>, I can really
appreciate some of the hardships and political and structural problems
you've hit.  I'm interested to learn more about all this, especially
if it helps Debian avoid having roadblocks between their efforts and
the efforts of all users of free software.

I'm not sure what you mean about "translator interference" or
"maintainer interference".  I'm not really clear on how you guys
actually work -- I understand the bit about collecting PO files and
the registry and the robot and all.  Do you ship those upstream to the
respective maintainers?  (Is this the problem of the "Unintegrated
POTs" in bash et al?)

How can Debian help the TP more, aside form the obvious collaboration
on translating PO files (l10n), working with upstream to i18'ize?

What things in particular have we done wrong?

- --
.....Adam Di Carlo....adam@onShore.com.....<URL:http://www.onShore.com/>

- --DAA03717.935048383/amsterdam.interport.net--

------- End of Forwarded Message

Reply to: