Shibboleth Developers

[Scott Cantor <>]

Marlena gave me a perhaps more effective way to describe the processing
model that we had in mind for this to begin with (probably because it
was hers to begin with), so let me see if this is clearer in
implementation terms:

Within the ARP classifications that CMU developed, there are basically
two types of ARPs, one type where the SHAR is a wildcarded expression,
and one type where the SHAR is an exact, literal name. In the former
case, the URL is always just '*', meaning any resource. Only the latter
case has a meaningful URL part.

The implication of that is that you can boil the steps down in finding
the right ARP to this:

Search the ARPs that specify exact SHAR names for the requesting SHAR
If found Then
        Find the best fit for URL
        If found Then
                Done, that's the right ARP to use
Search the ARPs that specify wildcarded SHARs and choose the best fit

As you can see, it's pretty simple. The other things that have been
discussed to support attribute filtering and the default/admin ARPs and
so forth all fit alongside that basic algorithm.

What I was describing about matching both the SHAR and URL at the same
time is technically correct, the result will be the same, but it makes
it sound less obviously easy.

-- Scott

------------------------------------------------------mace-shib-design-+
For list utilities, archives, subscribe, unsubscribe, etc. please visit the
ListProc web interface at 

    http://archives.internet2.edu/

------------------------------------------------------mace-shib-design--