Re: Sharing thoughts about the ddtp approaches
Le 16/10/2019 à 15:41, s3v a écrit :
> I believe automation requires a lot of work. We have:
> - many languages
> - many grammars
> - many syntactic rules
> - few contributors (none?)
> - no top-level rules (e.g. policies to advice maintainers to write descriptions
> using some schemas/variables).
> Furthermore, avoiding an automatic (and never reviewed!) mistake by DDTP is
> better than a bunch of automatic translation handled properly, IMHO.
> Maybe we should concentrate on "easily-to-translate" descriptions instead of
> automatic translations and addressing little problems
> for speeding up translation workflow would be a immediate solution for improving
> the "productivity".
Right. The thing is not addressing issues per language, but trying to
avoid duplicate work.
> Anyway I agree with kleptog: unfortunately DDTP doesn't have active contributors
> both for little and big actions.
>> a) python-* and python3-* have the same descs. We should be able to
>> translate both python and python3 translating one of them.
> python2 packages are going to be removed from sid/bullseye.
> Therefore, there is no need to translate them.
Well they should not be fetched then. Could not we filter them to avoid
>> b) linux-*-dbg: the string is similar, the translation should be
>> automated, optionally put in the reviewing string for safety
> dforsi from Italian Team wrote a very useful script for translating linux-*
> Note that this is on the way to "easily-to-translate" land, so it could not be
> what you are looking for.
How such script could be applied as we have not control on the database?
What is the good workflow?
> Kind regards and thanks for throwing a stone in the water.