> Have a look at Camomile:
>
> http://camomile.sourceforge.net/
>
> Generally speaking, though, I just always use string == UTF-8 string
> and avoid using some of the unsafe functions from the standard
> library, such as String.lowercase.

Yes I did (as if you read closer, I did mention). But being able to
use actual utf8 rather than manually encoding characters in the string
is what would be nice. Maybe I just need a script to preprocess
sources in utf8...