On Sat, 18 Oct 2003, Tuomo Valkonen wrote: > The problem with general mb stuff is the lack of a libc standard function > to go back one character in a string (M$ and Watcom libraries seem to have > an mbsdec function) and with general character sets I'm afraid this could > make the title-shortening algorithm too slow in case of long titles. (A > more complicated algorithm might be a possibility but I still don't like > scanning through the whole string all the time.) An alternative would be > to just use wchar_t internally, but I kind of prefer a nice encoding like > utf-8.
Ah, yes, I remember now. But using the Xutf8* functions is not a very good solution either. So... what about using UTF-8 internally but converting strings to the current locale when calling XmbDrawString? Or would that be too slow?
