Re: gettext "plural forms" in Turkish

[Bruno Haible]

Sertaç Ö. Yıldız wrote:
> > Following the same example, and translating to Turkish:
> > 
> >     “0 apples”      → “0 elma”
> >     “no apples”     → “elma yok”
> >     “1 apple”       → “1 elma”
> >     “2 apples”      → “2 elma”
> >     “123 apples”    → “123 elma”
> >     “the apple”     → “elma”
> >     “the apples”    → “elmalar” (!)

The ones without explicit numbers are different grammatical constructions;
therefore the relevant cases are:

> >     “0 apples”      → “0 elma”
> >     “1 apple”       → “1 elma”
> >     “2 apples”      → “2 elma”

Therefore nplurals=1 is appropriate.

Nilgün Belma Bugüner replied:
> In practice, already, it's so (from gettext.tr.po): 
>     "Plural-Forms: nplurals=2; plural=(n != 1);\n"
> 
> Subject lines:
>     #: src/msgl-check.c:384
>     #, c-format
>     msgid "but some messages have only one plural form"
>     msgid_plural "but some messages have only %lu plural forms"
>     msgstr[0] "...ama bazı iletiler tek çoğul biçem içeriyor"
>     msgstr[1] "...ama bazı iletiler sadece %lu çoğul biçem içeriyor"

A translator is free to choose a plural formula which fits better than
the default one. It appears that for Turkish, the simple one
    nplurals=1; plural=0;
fits for the majority of the cases, whereas some messages better use
    nplurals=2; plural=(n != 1);
if the translator wants to make the case n=1 look better.

Bruno