i18n repo is broken

[Matt Mackall]

On Thu, 2013-01-10 at 21:59 +0400, Nikolaj Sjujskij wrote:
> Den 2013-01-10 17:34:22 skrev Martin Schröder  
> <martin.schroeder at nerdluecht.de>:
> 
> > Am 10.01.2013 12:42, schrieb Nikolaj Sjujskij:
> >> Den 2013-01-02 10:22:42 skrev Matt Mackall <mpm at selenic.com>:
> >>
> >>> At this point, all work there should happen strictly on the stable
> >>> branch, in the i18n/ directory.
> >>  What is "this point"? Does this mean that no i18n work should be done
> >> in `default`?
> >
> > Yes. All translation work should be done on stable.
>   Er... So every new major version has to be translated during feature  
> freeze, isn't it? Seems a bit weird to me.

My model for how things should work is Wagner Bruna, who's kept the
Brazilian Portugese translation fully in sync for years now:

$ hg log -r 'keyword(i18n) and branch(default) and keyword(wagner) and
not merge()' | less

He stopped doing actual translation work on default back in 2010. His
changes to sync with 2.4 look like this:

$ hg diff -r '"2.4-rc"::2.4' pt_BR.po  | egrep '[+]msgstr ".+"' | wc
     60     420    2925

..which seems like a manageable amount of catch-up work.

Perhaps Wagner is not a realistic role-model. But there's also no
necessity or expectation that every translation catch up immediately
either.

-- 
Mathematics is the supreme nostalgia of our time.