Re: Gnome-games and scheme

[Danilo Segan <dsegan gmx net>]

Christian Rose wrote:

>Good catch. I didn't think of that, even though I know some German. Ok,
>now we have our counter-example. :)
>
>The question remains on how to solve this problem. Theoretically,
>translating 1000+ very similar messages like this could be automated and
>not necessarily very difficult, but the luggage in form of 1000+ extra
>messages in the po files, let alone the extra code for it, doesn't
>exactly make it appealing, nor realistic.
>  
>
Yes, I quite agree. It would be a maintainers' nightmare if it was to 
contain a thousand strings in one giant switch statement.

>Perhaps the message could be rewritten. I'm thinking of examples like:
>
>msgid "move the card \"%s\" onto the card \"%s\""
>
>This probably avoids gender problems etc, since it's the card that's
>always the subject and object now. Would it solve the problem for
>Serbian too?
>  
>
Well, actually, we do use this form in such cases, so it's probably 
unneccessary to restrict other translators who *do* have an option for a 
nicer and better string.

Also, this is just "acceptable" solution, not a really correct one (some 
would argue that even then the word should be tranformed in Serbian, but 
at least this is readable and doesn't sound *too* flaky).

Also, I'd be interested to know how are other translators who face 
similar dillemas approaching the issue. If we can generalize this 
enough, we might maybe come up with the solution similar to plural-forms 
in gettext.

What I know so far:
Serbian has 7 forms of the nouns and adjectives (not counting the 
difference of gender and plurality).
German has 4.
Russian (I believe, correct me if I'm wrong) has 6.
Croatian and Bosnian should have also 7.

The forms that are distinguished in Serbian are:
1. nominative
2. genitive
3. dative
4. accusative
5. vocative (hey, you)
6. locative
7. instrumental

I think many languages are actually having only a subset of these, but 
the forms are the same. Probably some language has other forms which are 
a superset of these (lets call this language the "most complicated" 
one). German uses the first 4, Russian does not use one (I think it 
doesn't use vocative), etc.

To perform translation in that way, programmers should code it the way 
that would work in the "most complicated2 language, because then others 
would be able to just ignore the unused forms, and replace them with 
other forms (using the mechanism similar to the "plural-forms" in gettext).

If a support in gettext would be developed, I think it would be nice to 
allow translators to "fix" these issues, but I don't see it coming any soon.

Ok, so these are just a bunch of random thoughs, hope someone will find 
something interesting in here.

Cheers,
Danilo