On 10/15/10 06:38, Andreas Gruenbacher wrote: > One of the key differences between quoting in gnulib and in git is that > gnulib > treats filenames as sequences of characters which may be multi-byte, while > git > treats them as sequences of bytes. > > I'm not sure about the reasoning behind ignoring multi-byte characters; maybe > Paul or Junio can enlighten us.
I can't speak for Jun, but I expect the answer is, "Don't worry about stupid encodings". If you use a reasonable encoding like UTF-8, git's quoting method works just like gnulib's does. I don't think diffutils (or git) needs to worry about what happens to users who insist on using those awful legacy encodings.
