[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-i18n-doc
Subject:    Re: Different formulas for plurals in Qt and KDE
From:       Chusslove Illich <caslav.ilic () gmx ! net>
Date:       2014-04-10 21:49:43
Message-ID: 201404102349.43262.caslav.ilic () gmx ! net
[Download RAW message or body]


>> [: Alexander Potashev :]
>> Mixing i18n() in tr() in one .po sounds wrong conceptually, not only at
>> the level of our problems with plurals.
>
> [: Albert Astals Cid :]
> This should not happen and if it's happening it should be fixed.
>
> At most it's happening in Summit, Chusslove is the guy that cares about
> it.

I don't think it is conceptually wrong to have messages from different
translation systems in the same PO file. In fact, PO format specifically
supports that, through format flags. (Here there is an inconsistency on
KI18n side, because every single message from an i18n() call should have the
format flag, and not like now only those that contain argument placeholders.
I'm thinking since some time how to best correct this.)

However, PO convention does assume that plural rules are same for all
messages. This holds by definition if a translation system actually use MO
files at runtime, through gettext calls underneath. If instead PO file is
converted to another runtime format, then it is the convertor (whatever
chain of commands for conversion) which should allow optional conversion of
plural forms.

For branch POs the issue is academic though, as Albert says, since there is
no reason for there to be a PO file containing both types of messages (a
framework uses either Ki18n or Qt, never both). Well, except for Scripty
forcing the plural form header on all POs (I'd simply kill that).

For summit POs, the "convertor" (to branch POs) is the posummit scatter
command, so there I'll prepare something little to put into
LANG/summit/messages.extras.summit for plural conversion.

-- 
Chusslove Illich (Часлав Илић)

["signature.asc" (application/pgp-signature)]

[prev in list] [next in list] [prev in thread] [next in thread]