Re: [dconf-editor] cleaning of translations, 2nd approach

[Mario Blättermann <mario blaettermann gmail com>]

2018-03-06 20:55 GMT+01:00 Rafal Luzynski <digitalfreak lingonborough com>:
> 6.03.2018 14:09 Arnaud Bonatti <arnaud bonatti gmail com> wrote:
> > [...]
> > 2018-03-06 13:47 UTC+01:00, Mario Blättermann <mario blaettermann gmail com>:
> > > Never touch po files, really never. […]
> >
> > I don’t find a way to express how sad I am to see that it looks
> > impossible for devs to help translators in other way than adding
> > things to translate; and for maintainers to have a global control of
> > the product they maintain. [...]
>
> I'm not sure how Damned Lies works but don't the project developers
> or maintainers have the power to contribute to any translation (in
> any language) for their projects?
>
> Back about your idea: I'm afraid it does not make much sense because
> your removed strings would be restored in the next translation update.
> Same about your updates which would be overwritten by the content
> of Damned Lies, no matter if the content is newer or older. And, finally,
> if the string is no longer used then it would be removed anyway so
> your patch does not do anything that Damned Lies would do, if used
> properly.

Damned Lies doesn't remove unused strings at all. They remain after
the underlying Vertimus tool has merged the po files with the latest
generated POT template (without any real changes in the git repo).
Translators can then decide to remove them, or not.

Unused strings can help to find a good translation for another string.
Well, nowadays the GUI tools like Poedit do this work using their
translation memory. But you must not expect that just everyone works
with specialized GUI editors. Maybe some (or some more, who knows...?)
people prefer simple text editors, and in that case the unused strings
can become very helpful.
> In the extreme cases, if the translation is too incomplete and
> unmaintained for too long one may consider removing the language
> from the project.

Who is "one" in this case? Who makes the decision that it's okay to
throw other's work away? A po file doesn't eat up that much disk
space, and as long as it doesn't break things, keep it as is.

In general, I'm in doubt about the benefit of your plans. A shrinking
of 3,3% of the tarball size is no real benefit. A (maybe) better
workflow for grep is no real benefit. With some clever grep calls, it
doesn't matter how big or "trashed" a file is. And last but not least,
you won't benefit from removing old and unmaintained po files, because
they don't cause extra work for you.

Best Regards,
Mario