Community technical support mailing list was retired 2010 and replaced with a professional technical support team. For assistance please contact: Pre-sales Technical support via email to sales@march-hare.com.
On 8/23/06, Gerhard Fiedler <lists at connectionbrazil.com> wrote:
> Yongwei Wu wrote:
>
> >>>> in my configuration, a utf-8 file added as Unicode ("add -ku") will be
> >>>> encoded as utf-16 on checkout.
> >>>
> >>> Yes, because you asked it to... that's what -ku means.
> >>>
> >>> If you don't want utf-16 then don't specify -ku.
> >>
> >> Oh... sorry :) I didn't RTFM.
> >>
> >> In this case my question is... it seems cvsnt doesn't support utf-8 then,
> >> right? utf-8 files should thus be added as binary, and won't get merged. Is
> >> this correct?
> >
> > Why not just adding it as plain text? UTF-8 is designed to work like
> > normal text files.
>
> I don't know what you call "normal text files", but with utf-8, a single
> character can be anywhere from one to four bytes. If with "normal text
> files" you mean single-byte ASCII, this doesn't work well. It works only as
> long as you use exclusively ASCII characters, because in this case, the
> byte encoding is in fact identical.
>
> What I'd like to know is whether cvsnt is designed to work with files
> encoded in utf-8 added as normal text files (-kt) -- for merging, for
> example.
>
> Gerhard
It may be the case that CVSNT is not *designed* for multi-byte
characters, but I do not feel any problems about it. I have stored
Chinese (double-byte characters) in CVS for years, and do not know of
any problems.
Hint: Merging is line based; and control characters and 0x24 (`$')
never appears in multi-byte character sequences.
Yongwei
--
Wu Yongwei
URL: http://wyw.dcweb.cn/