Oliver Tappe <zooey@xxxxxxxxxxxxxxx> wrote: > These tests have been done on R5, using a libiconv header that > declares iconv() > like this: > size_t iconv (iconv_t cd, const char* * inbuf, size_t *inbytesleft, > char* * > outbuf, size_t *outbytesleft); > > So, unless I miss something, my original idea of changing haiku's > iconv() > declaration is a bad one - as it could trigger problems for other > programs, at > least in C++ mode. Marcus obviously meant the other way around, but that doesn't help with POSIX conforming apps. Bye, Axel.