Re: Exporting gssapi context, take two

[Sam Hartman <hartmans@mit.edu>]

>>>>> "Kevin" == Kevin Coffman <kwc@citi.umich.edu> writes:

    Kevin> Sam, I understand (and can agree with) your aversion to
    Kevin> putting this in the glue layer, but aren't you asking for a
    Kevin> standardized glue api here?

    Kevin> I'm sure you don't want code that is dependent on a
    Kevin> specific glue implementation in the MIT gssapi code?  This
    Kevin> doesn't make it right, but there is already a pname_to_uid
    Kevin> function in the mechglue code.

I don't?  I'm not at all clear about that.  UNlike Java, the C
bindings do not specify the interface between the glue layer and the
mechanisms, so it seems like that interface needs to be glue layer
specific in some way.

But I never stopped to question the assumption that mechanism
implementations were bound to at most one glue layer.