[cvsnt] Re: why I can't get the filename include chinese character in output of "cvs history"

[Tony Hoyle]

Paul Russell wrote:

> 
> Some of our files have Japanese filenames and they store correctly, the
> CVSROOT\History file views correctly, and ViewCVS works fine too. Maybe
> everything is already in Unicode, although we did nothing special when

Unless MS have created a UTF8 OS I can't see how... Japanese needs more than
the 240-odd ANSI characters that you can generate.  There must be some
trickery going on to squeeze the extra characters in... I doubt very much
it's related to unicode directly.

> creating the files in Windows2000 and Windows98. This is with CVSNT

Win98 doesn't support anything to do with unicode, which seems to support
the assertion that it's not unicode.

Presumably you need the patched Japanese OS to see these files, and of
course nobody without these patches will be able to make sense of them...
this is the problem with non-unicode encodings, unfortunately.

The biggest problem with making international filenames and comments (aside
from having to rewrite the C library and write a wrapper around every
single Win32 API function... no small task in itself) is finding out what
codepage they're written in originally, which is nearly impossible - I'd
expect if cvs ever got that kind of support you wouldn't be able to keep
any existing repositories with non US-ASCII characters - it's not practical
to convert them.

Tony