Re: French character names in gucharmap

[Danilo Segan <danilo gnome org>]

Hi (again :) Noah,

Noah Levitt <nlevitt@columbia.edu> writes:
>
> I'm sure no one wants all the (15,000+) character names
> marked for translation. 
>

No one wants 15k strings more to translate, I'm positive about that ;)

My preference would be to create a separate unicode-data gettext
domain, which would not be tightly bound to Gucharmap, where you
would put all the character names and possibly all the other data
extracted from Unicode database (these names remind me of
UnicodeData.txt which was distributed in 3.x times, maybe it still is).

At the same time, you add translation to French as only current
translation, and put it on Translation Project, so that it is clear 
that it is not Gnome specific.  Others are free to pick it up from
these, if they want entire database or some parts translated, and
simply commit it.

I believe that most of the translators would choose to first
translate only strings pertaining to their own alphabet(s), but that
is good enough I think.  Unicode data doesn't seem to change too
often (i.e. everything already translated will probably stay the
same, there are only new additions in new releases), so everyone can
translate it gradually.

This has the benefit that it will certainly be no worse than the
current situation: if there's no translation, they would get English
originals.

Other benefit is that other projects would be able to reuse this if
they need/want to.

> Does anyone have an opinion on whether it's worth adding
> several hundred k to the the binary for the French character
> names? 
>
> If so, I'm leaning towards showing both the French and
> English names regardless of the user's locale. Does anyone
> have an opinion on that?

As I said, the best thing would be to separate this into another
gettext domain (I suggest "unicode-data", perhaps it should be
versioned as in "unicode-data-4.0"?), and simply use
dgettext("unicode-data",...) where you would use gettext(...).

This might present a performance penalty, so perhaps it's better to
bind "unicode-data" textdomain , and use regular gettext(); but, that 
depends on both your code, and gettext library.


Of course, this is simply my opinion, and others may think
differently. ;)


Cheers,
Danilo