[GRASS-translations] UTF-8 or ISO?

[Glynn Clements]

Martin Landa wrote:

> >  > Welcome all translators and coders.
> >  > I am newbie in translations and my question probably is simple one.
> >  > Which *.po files coding standard is recommended now?
> >  > UTF or ISO?
> >  > I continue polish translation (i.e. UTF-8 and ISO-8859-2).
> >
> >
> > UTF-8.
> >  It was an error when russian translation was switched from UTF-8 to KOI8-R.
> >  In most of the projects there is a tendency to switch to utf-8.
> >  Also, UTF-8 becomes default system locale. (For ru_RU at least.)
> 
> e.g. Po files for Czech translation are using iso-8859-2, but it
> doesn't mean that cs_CZ.UTF8 do not work. It is just question of po
> file encoding AFAIU.

GNU gettext uses UTF-8 for the .mo files, regardless of the encoding
used for the .po files, then converts to the locale's encoding at
run-time.

Some other gettext implementations store the data in the locale's
encoding, with no conversion at run-time. Such systems typically don't
use UTF-8 locales by default, and may not even offer them as an
option.

So, for GNU gettext the .po encoding doesn't matter, and a legacy
gettext will be better off with the locale's "historical" encoding
(e.g. ISO-8859-*, KOI8-R etc). Hence the preference to use the
historical encoding for .po files.

-- 
Glynn Clements <glynn at gclements.plus.com>