update ubuntu-docs errors

[Matthew East]

Hiya,

On Thu, Apr 9, 2009 at 1:53 PM, Alexey Balmashnov
<a.balmashnov at gmail.com> wrote:
> On Thu, Apr 9, 2009 at 2:27 PM, Matthew East <mdke at ubuntu.com> wrote:
> [snip]
>> The most effective and long-lasting way to fix these errors is if
>> corrected po files get uploaded directly to Launchpad. That way we get
>> the fixes for the next release too.
> [snip]
>
> As far as I understand, it is not entirely true.
>
> IIRC, fixes of Intrepid documentation were done way before Jaunty was
> open for translations. Despite that, strings that were fixed for
> Intrepid required another intervention in Jaunty.

If that's the case, it results from a failure to sync the translations
across from intrepid to jaunty in Rosetta, or in the ubuntu-docs
package. In either case it isn't intended behaviour and hopefully, it
won't happen again.

> Next point: unless there is a possibility of some sort of moderation
> (or lock or whatever) for particular pieces of translations in Rosetta
> (lets say, for documentation) it will be possible for translators to
> easily break translations of documentation.

Yes, that's true, of course. However there is no obvious reason for
translators to re-translate a string which already has a translation,
unless they want to improve on the existing translation. The number of
cases is likely to be very small.

> And again, about checks. Isn't it possible to implement sort of
> simplistic checks for translated documentation strings? Yep, I
> remember about some implementation of validation in gettext or
> something like that for XML-validation or something similar (I am not
> an expert), but still don't understand why it is not possible to
> prevent at least screwing up the XML tags? You do not need to validate
> a complete final XML document for that, especially taking into account
> that original documentation strings are coming to Rosetta from
> verified and validated sources, so that checking algorithm may look
> like:
> 1. Take original string, dump everything that is not tag
> 2. Take translated string, dump everything that is not tag
> 3. Compare results of 1. and 2. If they are not the same =>
> translation messes up tags.

This is a wishlist bug for Rosetta, I suppose. Only Danilo can tell
you whether it's likely to be capable of implementation before
Launchpad is open sourced, although of course after that, others might
be able to contribute to a possible fix.

-- 
Matthew East
http://www.mdke.org
gnupg pub 1024D/0E6B06FF