Shibboleth Developers

["Scott Cantor" <>]

> A browser plugin is an interesting theoretical solution to IdP
> Discovery, but if there's anything we've learned in the last ten years
> (since the introduction of Java applets) it's that any solution that
> depends on the client is doomed to failure (or at least problematic).

When you have a client problem, any solution that doesn't depend on the
client is doomed to failure by definition. I can tell you that Microsoft's
solution to this is to change the client. Coming from me it sounds insane,
but not coming from them.

You can certainly argue that the plugin idea is bad, and I probably agree.
But the implication to me is that the problem can't be solved, not that the
server can solve it.

> > Why not just have a webapp that does it?
> 
> I think I asked that awhile back and the answer was that the WAYF is
> stateless and should remain that way.  I agree with you though, Chad. 
> If there's a solution to IdP Discovery, it probably involves a webapp
> on the server.

As you've seen in my response to Chad, this has nothing to do with state.
There is no viable solution to this that depends on a server. There are a
lot of lousy solutions to it that do. And unfortunately all the current ones
fall into that category.

-- Scott