Am Samstag 14 November 2009 00:00:36 schrieb Jason Dusek:
I'm binding to `wcwidth` to determine the column widths of various Unicode characters. I noticed a lot of -- in fact all -- Chinese characters were being given widths of `-1` when of course they should have width `2`. This only showed up when I compiled my program though -- within GHCi, it never happened.
It seems that ghci calls setlocale(LC_ALL,"") or similar, while the compiled code doesn't. I've no idea why that would be, but dafis@linux-mkk1:~/Haskell/CafeTesting> cat locl.h void sloc(); dafis@linux-mkk1:~/Haskell/CafeTesting> cat locl.c #include <locale.h> #include "locl.h" void sloc(){ setlocale(LC_ALL,""); } main = do setloc (sequence_ . fmap (UTF8.putStrLn . uncurry fmt)) widths where... foreign import ccall unsafe "locl.h sloc" setloc :: IO () fixes it.