i18n: Starting Chinese Translation of Mercurial

[Benoit Boissinot]

On Sun, Jan 25, 2009 at 02:31:42PM +0100, Martin Geisler wrote:
> Mads Kiilerich <mads at kiilerich.com> writes:
> 
> > In an internal project we unfortunately couldn't extract translation
> > strings by static program analysis. Instead we had to extract the
> > strings on runtime, writing to a file when the translation engine met
> > an unknown string. That had the advantage of getting the most commonly
> > seen - and thus most important - messages extracted and translated
> > first.
> >
> > Perhaps the Mercurial i18n engine could be extended with an option to
> > do something similar, keeping track of how often each message
> > translation is actually used. Sorting the messages by that frequency
> > when translating would help translators to get started and focus on
> > the most "important" strings, make it "cheaper" to get good coverage,
> > and make it easier to establish a consistent terminology.
> >
> > Just an idea - I hope it could be seen as a way to improve i18n ;-)
> 
> It's a good idea, and I have already tried making mercurial.i18n.gettext
> count the number of messages it translates. Unfortunately there are a
> lot of messages that are translated without being used: the _(...)
> entries in mercurial.commands.table and in the extensions cmdtable's
> dictionaries.
> 
> Even with no extensions enabled this results in about 250 strings being
> translated on every startup.

On the other hand, command synopsis and help text should be high
priority even if they are not displayed often.

regards,

Benoit
-- 
:wq