>>Not that I like that too much - I was doing that in 2.6 for just two languages, and you simply can't keep something like that in code, you have to have a table with one column per language, and some code to generate .h files from it. Otherwise it becomes quite impossible to find your way around in an editor, once the number of strings surpasses a few screenfuls.
>
>As you observe, it is a nice theoretical solution but far less so in practise.
>And I wonder how easy it is to find an error in the (now hidden from view) "logic" within the 'code'?

And you can do quite weird things... I once had a really messy piece of special manually done report, where I had a lot of fwrite(h,SomeLengthyPrefixExpression+lcText) repeated all over the place. So I did this:

#define push fwrite(h,SomeLengthyPrefixExpression+

push "text 1")
push "text 2")

It reduced the total length of the code by nearly a third, worked great and confused me completely next time I looked at the code.