Re: Emacs i18n

[Paul Eggert <eggert@cs.ucla.edu>]

Richard Stallman wrote:

> If this does not work, why not?

Thanks for explaining the -masc, -fem, -neut part. I'm afraid, though, that I 
still don't fully understand the proposal. It sounds like it is a redesign of 
what GNU gettext does, but I don't see any advantage over GNU gettext.

> In the example that was sent, I see code that tests for certain kinds
> of numbers.  But since I don't know the language that that is written
> in, the mathematical conditions are the only part I understand.  I
> don't see what it will _do_ in each of those conditions.  I presume it
> selects the appropriate suffix for the number, but I don't follow how
> it does so.

The GNU gettext translation code doesn't know anything about suffixes. All it 
knows is that if n%10==1 && n%100!=11 then it should use msgstr[0], else if 
n%10>=2 && n%10<=4 && (n%100<10 || n%100>=20) then it should use msgstr[1], else 
it should use msgstr[2]. The translations themselves are string formats that 
already have the proper suffixes, and GNU gettext simply copies those suffixes.

This is a simple scheme that does not attempt to solve the problem of generating 
idiomatic phrases for numbers (e.g., "twenty-four bytes" in English, "двадцать 
четыре байта" in Russian). All it solves is the problem of generating phrases 
containing numerals (e.g., "24 bytes" in English, "23 байта" in Russian), as 
these are the sorts of phrases that printf formats can generate. In practice, 
this is good enough.