Re: Bug#457946: aptitude: Incorrect plural in commandline status

[Daniel Burrows <dburrows@debian.org>]

On Fri, Dec 28, 2007 at 08:01:16PM +0100, Christian Perrier <bubulle@debian.org> was heard to say:
> Quoting Daniel Burrows (dburrows@debian.org):
> >   Yes; my question is, is ngettext going to do the right thing for
> > lists in all languages?  In English, we conjugate based on the total
> > number of items in the list (although both "there are one foo and three
> > bars" and "there is one foo and three bars" sound awkward to me).  I
> > wouldn't bet on this convention being the same in all languages, and I
> > figure Christian might have a better idea than me, since he's an expert
> > on the subject of translation.
> 
> 
> Well, if I understand ngettext correctly, it should do The Right
> Thing, ie use a plural-form in the POT file so that languages with
> various ways to conjugate/enumerate things can put as many variants as
> possible.

  My concern is that the number we're providing might not be the correct
one in all languages.  If we want to say:

    there are 5 X's, 3 Y's, and 1 Z

  we provide the number 5+3+1=9 to ngettext().  That means that however
many templates there are in the .po file, there's no way for a language
that cares only about the first number in the list to provide a correct
translation.

  Probably I'm thinking about this too hard and I should just apply the
proposed patch, and wait for translators to complain that it's wrong.  :-)

  Daniel