Re: Localisation Guide

[Dafydd Harries <daf muse 19inch net>]

Ar 04/01/2004 am 23:09, ysgrifennodd Telsa Gwynne:
> > Through most of the part about editing the PO file, you assume that
> > the translator is editing the file as pure text. It'd be better to
> > recommend, from the outset, to use a PO-editor or a PO-aware editor
> > such as Emacs, Gtranslator, Kbabel or PoEdit.
> 
> You'd suggest moving the list of editors we know to work well to
> the start, then? We do include such a list. We just put it after
> all the stuff about the files. 
> 
> Or that we should write more on the ways of editing which are
> not "text editor on a .po file"? The reason we wrote more about
> editing po files is that that is what we know most about.

I think this is the main reason we didn't emphasise editors which
specifically support .po files: we haven't used such tools ourselves.
Most of the Welsh translations were done with a simple text editor and
one or two macros. (Specifically: one to find the next untranslated
string and one to update the modification time header.) Virtually all
the other translations came from the web-based interface, and they would
have been tidied up with a text editor before being committed.

I also agree with Christian's point though: we should emphasise the
simple, basic ways of doing things because those are the methods that
everybody will be able to use.

If people want to suggest links that could be added to point to things
like PO mode or Gtranslator, or documentation for them, that would be
great. Also, if you have macros for your favourite text editor that
helps things go smoothly, or tricks which help ease the burden, that
would be useful.

> > 	"TRUE" and "FALSE" show up in gtk+ and in Gconf and in a lot
> > 	of files generated by Glade. (These will have a name ending in
> > 	.glade.) Do not translate them. Programs expect to see them
> > 	and will be confused if they don't.
> > 
> > Do they expect the substrings "TRUE" and "FALSE" inside the translated
> > strings? Not the integer values 1 or 0 in some variable?

I think the thing is that the literal strings "TRUE" and "FALSE" are
expected not within the translated string, but elsewhere. Much like with
GConf key descriptions: the translated description is not parsed for the
sample values, but the sample values must be verbatim because the
translated value would not be recognised.

> > You don't mention GConf key descriptions. You should add a note that
> > the possible key values are often listed in the description string,
> > and those values should not be translated.
> 
> We should.

I've committed a new revision (0.7). It mentions GConf key descriptions,
and adds a link to the translation system that Funda mentioned.

-- 
Dafydd