Unicode and APIs

[Nicolas Williams]

On Wed, Sep 19, 2007 at 04:54:11PM -0500, John Hascall wrote:
> 
> 
> > I think the challenge will be credential caches,
> > keytabs, replay caches, etc.  Those are resources which are shared with
> > other Kerberos implementations that will not necessarily be happy if the
> > character sets changes.
> > 
> > I can see the possibility of issues where tickets for the same service
> > get stored under two different names, or keytab entries that cannot be
> > found, or replays that are permitted between applications which use
> > different character sets.
> 
> I'm beginning to suspect that taking the
> accumulated wisdom of the last 15 years and
> just biting the bullet and starting on 'Kerberos V6'
> is the one true way forward...

You'd still have to migrate existing deployments.

And you'd still have to write the code.

The problem here is not so much the protocol, but the operating systems
on which the protocol is being used *AND* the bad habits (just-use-8)
that have been bred over the years.