Processing .k5login (another patch)

[Roland C. Dowdeswell]

On Fri, Sep 03, 2010 at 12:17:40AM +0100, Roland C. Dowdeswell wrote:
>

> Defining types to be:
> 
> 	NONE		nothing.
> 	FILE		simple file from argument
> 	ANAME2LNAME	use krb5_aname_to_lname() to see if it matches
> 	KDB		use a KDB looking for an entry of the form
> 				principal\0luser
> 			with a key of ``1'' indicating yes
> 
> with this, the current behaviour would be defined to be:
> 
> 	k5login = FILE:%h/.k5login
> 	k5login = ANAME2LNAME
> 
> (roughly, because currently it looks like if ~/.k5login exists but is
> not owned by the right people the ANAME2LNAME is short-circuited.)

I didn't make it clear in this e-mail: I think that something like
ANAME2LNAME as a type is required to subsume current semantics and
I think that it would be better to be explicit about how it is
evaluated rather than just falling back on ANAME2LNAME lookups in
some of the cases where the files are not found (the current
behaviour.)

--
    Roland Dowdeswell                      http://Imrryr.ORG/~elric/