On Feb 1, 2007, at 3:33 PM, Scott Battaglia wrote:

> Would your DartmouthUserPassCred2PrincipalResolver be able to  
> return a principal for the types of users that the new  
> CredentialsToPrincipalResolver would be able to handle?  If it  
> doesn't recognize those users and would return null, then the chain  
> would continue until it found one that did not return null.

I *think* it should fail, the backend data source it would be trying  
to pull from is very different, and there would be no overlap of  
users. I wasn't sure what would happen if the first UserPass resolver  
failed, but sounds like it will do what we need.

> I am looking at creating a more optimized algorithm (an alternative  
> AuthenticationManager) that would allow you to specify a one-to-one  
> mapping of authentication handlers and  
> CredentialsToPrincipalResolvers.  We do it the way we do now as it  
> covers both scenarios where people have the mapping (its just less  
> efficient) and people don't have the mapping.
>
> Thoughts?

Might make it simpler, I can't ever see a need for an AuthHanlder to  
work with more than one Resolver. One C2PResolver should be able to  
support mulitple Handlers though. It is more to configure, but I  
don't know if it makes it easier in the actual code.

As long as the fall-through is fast and well-defined, I don't see a  
strong preference. One question would be what happens with 25 auth  
handlers and say 20 C2PResolvers?

Steve
_______________________________________________
Yale CAS mailing list
[email protected]
http://tp.its.yale.edu/mailman/listinfo/cas

Reply via email to