Re: UTF-8 and ISO8859-1 interoperability problem

[Mark D. Baushke]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi Joonas,

I have noticed in the past that HFS filesystems do odd things to UTF-8
characters in filenames. A web search turns up something that the
TWiki Web-based collaboration platform ran into similar problems. See
the URL:

  http://twiki.org/cgi-bin/view/Codev/MacOSXFilesystemEncodingWithI18N

for more information.

Right now the cvshome.org version of cvs does not do anything special
with filename characters treating them as opaque 8bit bytes.

You may wish to check to see if the cvsnt.org (or possible the
tortoisecvs.org project) folks have already dealt with the problem of
UTF-8 in filenames. I remember sing UTF8 references in the CvsNT
client/server section. If they do have the problem already handled, we
would want to probably pull in patches that use the same extensions they
are using to address this problem.

        -- Mark

Joonas Lehtinen <jole@jole.fi> writes:

> We have CVS server running Linux and using ISO8859-1 as filename
> encoding (on ext2 filesystem). When I try to update directories
> containing files with high-bit characters on UTF-8 system (MacOS X
> 10.3), it leads to this problem:
> 
> $ cvs update
> $ cvs update
> cvs server: Updating .
> cvs update: cannot write ./Test-���: Invalid argument
> 
> My guess is that cvs tries to create a new file using ISO8859-1
> representation of the file on UTF-8 system, that fools the system to
> believe that filename contains UTF-8 multibyte characters instead of
> just ISO8859-1 characters.
> 
> Is there any workarounds for this? I have hundreds of files I must
> checkout, but can the checkout hangs when it encounters any file with
> high-bit set (�, �, ...). Is there a way to force cvs just to ignore
> problems and continue checking out the rest of the files?
> 
> Tried cvs versions: 1.10 (came with MacOS) and 1.11.5 (binary from
> fink project)
> 
> -- Joonas
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (FreeBSD)

iD8DBQE/1MAo3x41pRYZE/gRAg69AKDATRKLTQ0DrhTO3ihvKOnYRXWbmACZAQte
Jw2q8gRH9XPmqXKpboMf7HM=
=YVqO
-----END PGP SIGNATURE-----