Shibboleth Developers

["RL 'Bob' Morgan" <>]

> > And yes, this comment applies to existing text in 
> > shib-arch-04, specifically the last sentence of section 
> > 5.6.4, which I suppose is more or less the same as the result 
> > of today's debate.  I think I recall saying this a few months 
> > ago, though of course I probably also volunteered to write 
> > new text regarding this which of course I haven't.
> 
> I don't recall a specific motion to remove all of the detail surrounding
> ARPs, which is what this would need to be, I think. It's a lot more than
> just a sentence. I don't see any reason to describe anything detailed
> about them at all if they aren't explicitly going to be consistent
> across AAs.

A lot of specs provide info about example implementations, eg as
pseudocode (RFC 1510 for example), as part of expressing how protocol
messages are to be processed, sometimes saying "any implementation that
produces results consistent with this is fine".  I think that's what we're
doing here.  The material in 5.6.3 is all useful and fine to be in the
arch doc IMHO, to explain the kind of functionality we think should be in
a useful AA (I'd make a few tweaks to say "a policy might be expressed as
..." etc).  Some of the text in 5.6.4 assumes too much, as I said before,
but the middle paragraph (saying you should think about precedence among
policy expressions) seems to me to be useful advice, regardless of how
policy is managed, and I wouldn't want to lose it.

> I'm open to moving this to a separate document, I guess. What do others
> think?

I don't think this is necessary, if we just modify the existing material 
to emphasize we're illustrating desirable AA capabilities, not mandating 
an implementation.

 - RL "Bob"


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

    http://archives.internet2.edu/

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