Porting Heimdal's libkafs to MIT Kerberos

[Steven Michaud]

Thanks for the quick answer.

> I just don't see anything you gain from folding it into the main
> Kerberos distribution.

Until I'd read Darren Tucker's note
(http://marc.theaimsgroup.com/?l=openssh-unix-dev&m=107363804131871&w=2),
and also Ken Horenstein's, I'd have been inclined to agree.  Now I see
that you gain the ability to use krb5-config in your configure script.

The author of a libkafs port would (at the very least) have to detect
and alter an existing MIT krb5-config script.  But maybe this could be
done without actually patching the MIT distribution.  For example, you
could capture the output of the existing krb5-config script, then
rewrite it to add flags for your libkafs port (you could even "borrow"
most of the code to do this from a recent MIT Kerberos distribution).
Is the kind of solution you would suggest?

In any case, I think it's safe to tell the OpenSSH folks not to expect
libkafs in MIT Kerberos anytime soon :-)

On Fri, 9 Jan 2004, Sam Hartman wrote:

> >>>>> "Steven" == Steven Michaud <smichaud at pobox.com> writes:
>
>     Steven> Is MIT interested in porting Heimdal's libkafs to MIT
>     Steven> Kerberos 5, or in incorporating someone else's port (if
>     Steven> one gets written) into the standard distribution(s)?
>
>
> Mit is very reluctant to treat AFS as a special application.  There is
> support for AFS in src/appl/bsd/login.krb5.  I think some of us view
> that support as a bug.
>
> MIT certainly does not have resources to port libkafs.  MIT would be
> very unlikely to pull libkafs into the distribution and would
> discourage others from doing so.
>
> That said, I think libkafs is probably the right way to go for AFS
> support and would like to see a version of libkafs that worked with
> MIT Kerberos.  I just don't see anything you gain from folding it into
> the main Kerberos distribution.
>
>
>