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

Re: String freeze (was Re: Debian Debconf Translation: Compromise)



On Tue, Jan 13, 2004 at 07:18:32AM +0100, Christian Perrier wrote:
> Quoting Denis Barbier (barbier@linuxfr.org):
> > Sounds fair.  I will go through the BTS and collect l10n related
> > bugreports so that they can be worked on.  If there are other people
> 
> How the hell will you do this as there is no "i18n" tag in the BTS
> (wild wide ranting)?

OK, enough already please.

If the i18n tag were added to the BTS, would there immediately be
requests for i18n-fr, i18n-de, etc. (or l10n-, more sensibly) tags?

If the answer to this question is "no, never", then it can be added
reasonably easily, but switching to a non-tag-based system would then be
tedious (you'd have to migrate all existing bugs, arrange for the old
tag to be known but refused in much the same way as the fixed severity
is now handled, and so on).

If the answer to this question is "yes" or "maybe", then a tag is the
wrong answer. I don't want to set ourselves up for a tag explosion when
that isn't the right way to do it. Instead, we'd want a new piece of
metadata, say "Language"; there'd be a new pseudo-header so that you
could say "Language: fr, de", and a couple of new control commands; we'd
have to think about whether we wanted to restrict the set of allowed
languages or whether you just want to fix it up after the fact when
people say "Language: french" instead of "Language: fr"; and we'd have
to implement searching for it. It's only recently that it's become
possible at all for us to add new metadata, following an internal change
in a file format, and we don't yet have any means to extend the indexes
so that we can search on new metadata. This can probably be solved,
though.

In this latter model, you'd also have to think about how to represent
bugs that report some problem in internationalization (i.e. not relating
to any specific language) as opposed to localization.

> Several of us are using tricks (such as [intl:xx] markers in BR
> subjects), but that's not really satisfactory as all i18n related bug
> reports are far from being marked like this.

That's not a very persuasive comment, you know; the bug reports aren't
tagged, either ...

There is one reason why tags are better than subject line hacking, and
that's that you can currently only search on subject lines within a
single (source) package, but you can search on tags throughout the whole
system. I should fix that, I guess.

Cheers,

-- 
Colin Watson                                  [cjwatson@flatline.org.uk]



Reply to: