Rosetta-Feedback - UDS Prague

Danilo Šegan danilo at canonical.com
Wed Jun 18 15:43:59 BST 2008


Hi Kenneth,

Today at 12:56, Kenneth Nielsen wrote:

>> I think you are missing one important point.  A "reviewer" can also
>> submit translations without waiting for them to be reviewed.
>
> No

Sorry, but yes :) Whether that's desired is a different topic.

> I.e. by reviewing someone's translations, you are aiming for a new
>> 'trusted' translator as well.
>
> Not if we can't provide feedback and make them make the corrections
> themselves.

I already acknowledged that we are missing commenting feature and that
it's important for reviewer's work.  I also mentioned how external
communication is still essential for running a translation team in LP
(due to lack of such commenting feature).

>>  So, now it'll be two guys who can
>> translate directly, and if that doesn't speed you up long-term, I
>> don't know what will.
>
> I think you are missing an important point here. There are _many_ upstream
> translators/translation teams that consider reviewing translations, even
> those done by seasoned translators, as a integral part of the translation

Actually, I am quite aware of that.  And I am doing some improvements to
enable reviewers to submit only suggestions (which is what they can't
do right now).

> process, that is absolutely necessary to reach a high quality output. E.g.
> _all_ translations submitted to the GNOME SVN for the Danish language has
> been reviewed, indenpendently of the translator. We don't want to comprimise
> our standards for quality in Ubuntu, hence what we need is a way to quickly
> review, _not_ "correct", a translation, because if we correct them ourselves
> then translator will not learn anything and the reviewers will have to keep
> correcting the same things.

You do make a strong case here.  I am thinking out loud, but if we
modify the translator evaluation page to group submissions by
dates, I think we'd have a good enough approximation to your 'point
diff' approach.

The mechanism to comment on such sets of suggestions would be very
useful, indeed, but I think intially just having them per single page
which you could copy and paste into email should be a big help.

> What we do when we review is to read through the translations, commenting
> only on the one that needs commenting, and only in as much detail as is
> required for the individual string. This can sometimes only be a single word
> or sentence "Typo", "Reformulate to avoid english sentence structure",
> "misgid has plural" or sometimes it can be a long explanation of some
> preferred terminology or policy. This all means that I can review and
> provide feedback for translations, as fast as I can read, write and delete
> text in a text editor and send an email, and _that_ is what I am looking
> for. My suggested "point-diff" approach will allow for that, in parallel
> with anything else you guys might think up as the main approach in the
> web-interface.

Actually, I am sorry to say that I haven't heard a single comment
about the existing evaluation view.  The ideas from this thread can
make it much more useful, and I'd be happy to spend some more time
thinking about this, and later working on it.

However, don't put up your hopes too high: it will take a while until
I find time to work on this, especially now that it's just Jeroen and
me working on LP Translations.

If anyone feels like working on Launchpad Translations, check out

  http://webapps.ubuntu.com/employment/canonical_LTSE/

:)

Cheers,
Danilo



More information about the ubuntu-translators mailing list