Re: gettext locale environment variable documentation

[Bruno Haible]

Hi Karl,

> Back on the new user chapter in the gettext documentation.  The new text
> says:
> 
>     A locale name usually has the form @address@hidden@var{CC}}.  Here
>     @address@hidden is an @w{ISO 3166} two-letter country code.
> 
> What about languages which don't have two-letter codes?

Two sentences later, there is a reference to the appendix "Language Codes"
of the same manual. I think this answers it.

> Does gettext support those?

It's an important principle in Unix internationalization that the list of
locales is extensible. There is no central list of the "supported" locales
in the system (such as in Microsoft software). Everyone is free to define
of locale of his own by
  - choosing a locale name that does not collide with the existing ones,
  - creating a locale description file for glibc, and using "localedef",
  - creating a symlink in the X11 directories,
  - creating PO files for various packages,
  - ...

> I see that `locales_with_principal_territory' in
> msginit.c has entries for various three-letter codes (starting with
> ace_ID), so I imagine they are ok.  In which case I suggest the doc do
> something other than flat-out state that it is a two-letter code, e.g.,
>   ... an @w{ISO 639} language code, which is usually two letters but
>   may be three ...

I think the appendix "Language Codes" explains it all.

> By the way, where did that principal_territory table come from?

I compiled it from information found on www.ethnologue.com.

If you want to dive deeper into the matter, you can take a look at [1].

Bruno


[1] 
http://en.wikibooks.org/wiki/FOSS_Open_Standards/Standards_and_Internationalization/Localization_of_Software