Re: Proposal for declinations in gettext

[Danilo Segan <dsegan gmx net>]

Hi,

Edward H Trager wrote:

>Why not just use another index for plural forms instead of "msgid_plural"?
>  
>
msgid and msgid_plural are used for "default" strings (English) which 
are to be translated, and translations which can contain arbitrary 
number of plural-forms are ATM described as:
msgstr[0] "Translation of plural form 0"
msgstr[1] "Translation of plural form 1"
...


>While in many languages the pronounciation (and thus the spelling) differs
>only between "singular"  and "plural", there *ARE* also languages that
>distinguish *MORE* forms, such as "singular", "two of an item", "plural
>(being three or more of an item)".  I believe the maximum in a
>known/studied human language is to distinguish *FIVE* different forms.
>Unfortunately, I would have to research what language(s) are like that,
>since I no longer remember.
>
Actually, I am aware of that (Serbian uses three forms: numbers ending 
with 1, except those ending with 11; numbers ending with 2--4 except 
those ending with 12--14; and all other numbers).

This cannot solve the problem at hand, because it would put more 
restrictions than needed:
-- programmer would have to use ngettext with the correct number of 
declination
-- languages with different number of declinations would fail
-- combining plural forms and declinations would not be easily achieved 
(because they're independent of each other mostly, so the total number 
of combinations is number of plural forms times number of declinations).


Cheers,
Danilo