JavaScript i18n support can fail for languages with two plural forms

[Ramiro Morales]

If, as exemplified in the official GNU gettext ​documentation, the plural form specification headers

"Plural-Forms: nplurals=2; plural=n != 1;\n"

or

"Plural-Forms: nplurals=2; plural=n>1;\n"

are used instead of

"Plural-Forms: nplurals=2; plural=n != 1 ? 1 : 0;\n"

or

"Plural-Forms: nplurals=2; plural=n>1 ? 1 : 0;\n"

respectively in a (Django or application) djangojs.po catalog, the ngettext() function fails because the first ones, being C language expressions, rely on the C feature that boolean expressions have to value zero or one. In JavaScript, a boolean in the context of an array index doesn't evaluate to zero or one.

See ​http://groups.google.com/group/django-users/browse_frm/thread/12f39fe090d39e78?hl=en for the original report by Julien.

This could affect JS i18n for languages that use two-forms pluralization rules, i.e.: Danish, Dutch, English, Faroese, German, Norwegian, Swedish, Estonian, Finnish, Greek, Hebrew, Italian, Portuguese, Spanish, Esperanto, French, Brazilian Portuguese (see the GNU gettext documentation linked above.)

The attached patch modifies the pluralidx() function to work on both scenarios.

[Malcolm Tredinnick]

Thanks, Ramiro. Saved me from having to write this. I agree we need to handle both forms, because the boolean form will be copied from other, similar files.

[Malcolm Tredinnick]

(In [7356]) Fixed #6864 -- Handle Javascript i18n when the plural form expression returns
True or False (so both booleans and integers are allowed). Thanks, Ramiro Morales.