Maniac wrote: > > Robert Wittams wrote: > >> It would of course be another option. But not a particularly good one. >> It means that when you change the template, someone has to merge the >> changes into all the other languages. You will almost certainly end up >> with your logic out of sync in different templates. >> Nightmare. >> >> > Unfortunately for some languages it's the only option. > > Taking the example of pluralizations of quantities it won't work for > Russian. Words depending on their form have different alforithms of > adjustment for numbers. Typical words have 3-4 different forms for > different numbers. And I can't tell exactly how many of those forms > exist but it's like some dozens of them :-). So simple singular/plural > just won't cut it :-). >
That isn't how ngettext works. The simple singular/plural is for the "C" locale, ie US English. A language can, AFAIK, provide many different substitutions depending on the specific number you are pluralising for. If you have a bit of text that is impossible to express in this way, of course you need to create your own mechanisms. But this should certainly not be the default, and is a far worse option in most cases. The only "complete" solution is to duplicate your entire application for every single language, so you can fit nuances of national character into every area. Not particularly maintainable. gettext seems like a sensible trade off at the current time.
