New i18n interface for KDE 4, second try

[Nicolas Goutte]

On Sunday 30 October 2005 21:12, Chusslove Illich wrote:
> >> [: Chusslove Illich :]
> >> Performance is really not an issue here, as we're talking about GUI
> >> stuff.
> >
> > [: Nicolas Goutte :]
> > Well, we have complex dialogs too for example. They should avoid to need
> > seconds more to load on a not-so-fast computer.
(...)
> > Keeping the interface simple means that you can have less surprises.
> > Let QString do what it can do well and put in the special class only
> > what is needed for a correct translation.
>
> I don't quite understand what posibillity you refer to here, but I do try
> to keep things as simple as possible. Providing that it works, and aside
> from that missed call names problem, what particulary is complex about
> this template solution, from the point of usage?

I meant that if QString::number does the job, we do not need a wrapper around 
it. (What the proposed qarg are, even if the shown code used an indirect way, 
instead of calling QString::number directly.)

>
> On the other hand, I think the undefined behaviour of arg() methods and
> need to keep an eye when multiple arg() should be used, shows that current
> QString solution already has some nasty surprises set up.


>
> > Especially that I suppose that numbers do not need really a special
> > handling for scripting.
>
> The numbers themselves might not require changing, but something based on
> them might. For example, those extra plural forms that we were talking
> about. Meaning that numbers as well must be supplied for scripting. And
> anyway, making some arguments available and some not, would just create
> inconsistency and confusion.

Sure, but counts can only be integers (Ooh, well they could be 64 bit ints 
too).

Sorry, if I was not precise enough. I was not menaing counts, I was really 
meaning numbers, like for the example (X: %1 Y: %2) which was given.
I meant that for such a use QString::number is enough (and the QString result 
can be given as parameter to the translation class.)

On the other side, I am just remembering that one of the original arguments 
for translation scriptings were phrases with multiple plurals, so that is 
perhaps a problem indeed, as the multiple counts must be given.

Have a nice day!