svn commit: samba r15209 - in trunk/source: libads libsmb

[Jeremy Allison]

On Mon, Apr 24, 2006 at 08:02:24PM +0200, Love wrote:
> 
> jra at samba.org writes:
> 
> > Author: jra
> > Date: 2006-04-24 15:57:51 +0000 (Mon, 24 Apr 2006)
> > New Revision: 15209
> >
> > WebSVN: http://websvn.samba.org/cgi-bin/viewcvs.cgi?view=rev&root=samba&rev=15209
> >
> > Log:
> > Add wrapper functions smb_krb5_parse_name, smb_krb5_unparse_name,
> > smb_krb5_parse_name_norealm_conv that pull/push from unix charset
> > to utf8 (which krb5 uses on the wire). This should fix issues when
> > the unix charset is not compatible with or set to utf8.
> 
> This is accually more fun. Kerberos uses ASN.1's GeneralString (which is
> mostly useless and ends up being ascii). Kerberos extentions will properly
> support i18n, but that in the future and given the blazing speed of krb-wg,
> it will arrive in about 1-10 years.
> 
> That said, I've tried to find out what ms sends, and that seems to
> something close to ldap normalized utf8 or raw utf8.
> 
> Watch out assuming that the kerberos api is a utf8, MIT folks seems to
> think its not but rather a locale specific interface, but I can't speak
> form them, ask the krb-dev list for what they think.

Thanks for that. I haven't seen any locale specific or multibyte
code in the MIT krb5, so as long as they treat it as plain byte
strings and the Windows krb5 code uses utf8 it should work between
the two (fingers crossed :-).

I'll take another look inside the MIT code but I don't recall
any locale specific stuff or multibyte string calls at all.

Jeremy.