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

Re: multistrap: [INTL:pt] Repeated errors in PO submissions



notfound 670715 2.1.18
quit

On Sat, 28 Apr 2012 13:35:46 +0100
Traduz - Portuguese Translation Team <traduz@debianpt.org> wrote:

> Package: multistrap
> Version: 2.1.18
> Tags: l10n, patch
> Severity: wishlist
> 
> Updated Portuguese translation for multistrap's PROGRAM messages.

... using an outdated PO file from who knows where.

pt: 111 translated messages, 1 fuzzy translation, 6 untranslated
messages.

No different to the existing PO figures.

> Translator: Pedro Ribeiro <p.m42.ribeiro@gmail.com>
> Feel free to use it.

No thanks.

> For translation updates please contact 'Last Translator' or the
> Portuguese Translation Team <traduz _at_ debianpt.org>.
> 
> Attaching the correct file for program's messages.

It is the PO file for the program messages but for an old version of
the program, not the PO file submitted in the request nor the PO file
sent attached to the previous reply.

The correct PO file, as attached here and previously, contains new
messages, one of which is:

#: ../multistrap:1492
#, perl-format
msgid "Apt preferences file to use: '%s'\n"
msgstr ""

That does NOT appear in the pt.po you submitted for this bug. The pt.po
you submitted already exists in the package, so closing this bug.

Attaching the right version AGAIN.

For the benefit of the debian-i18n list, this is the latest in a series
of failures with the Portuguese l10n support for multistrap. Things
began with translators slavishly following the templates and filing bugs
for debconf messages (which don't exist) or program strings when manpage
strings had been submitted. It isn't just Portuguese, other languages
have had some problems but each incidence is just as much work at this
end.

This is the last time I'm messing around with this. If the next pt.po
file is invalid, wrongly described, incomplete or has any errors at
all, I'm permanently dropping Portuguese as a supported translation for
multistrap and I'll refuse future requests for Portuguese translation
updates, not just for multistrap but quite possibly for my other
packages in Debian.

I'm beginning to reconsider using debian-i18n for anything *except*
debconf translation. It does not make it any easier to consider giving
translators direct access to package updates. It certainly does not
encourage me to respect the "contributions" of l10n teams or support
requests to NMU translation changes outside D-I or debconf. There are
individuals whom I would trust but if my work on multistrap,
emdebian-grip, drivel, svn-buildpackage and emdebian-crush are any
indication, the majority of translation contributors are not capable of
handling multiple modes of package translation (debconf, program &
manpages). I no longer feel that I can push for wider translator access
to packages within Debian. There is simply no effective QA. Any attempt
to revise the TDeb proposal would simply aggravate maintainers by
allowing poor quality updates which cause regressions in l10n support.
(If a manpage translation is submitted as a program translation, the
program translation regress to 0 translated strings - that is what I
consider an unacceptable quality l10n submission.) The people pushing
for this kind of support would share some of the blame for the
resulting incompetence.

Sadly, there seem to be few other sources of translators so I may
simply stop supporting i18n in my packages. Preparing i18n support is a
lot of work upstream and for the maintainer - if that work is
disrespected by the l10n teams by submitting poor quality "updates", it
isn't worth my time doing the i18n work. It is immeasurably easier to
write new code which does not support i18n than to write code which
does. Translation teams would be well advised to remember this.

Apologies won't change anything here - there needs to be structural
change in how the l10n teams work to guarantee quality submissions into
Debian. The current system just doesn't function effectively. There
*must* be a barrier between the translators and maintainers, some layer
which can make some assurance of quality by rejecting updates like the
ones I've had to handle.

-- 


Neil Williams
=============
http://www.linux.codehelp.co.uk/

Attachment: pt.po
Description: Binary data

Attachment: pgpoI1Akcb0Z_.pgp
Description: PGP signature


Reply to: