[OpenID] XRDS RP discovery when dynamic pages allow logins?

[SitG Admin]

>According to the OpenID 2.0 spec (as I read it), the RP discovery 
>feature requires that the return_to URL be found in the XRDS doc 
>published by the RP at the realm URL.  However, some sites, such as 
>blogs, allow logging in on virtually every page on the site 
>(thousands).  How should this be handled in the XRDS document since 
>it can't be practical to include thousands of potential return_to 
>URLs in the XRDS doc?

I've always thought of that as being part of the transparent (so it 
never bothered me to not notice this happening) redirection process; 
the *site* (Relying Party) keeps track of where a user was (URL#1), 
sends them to their OP (URL#2) with a return_to of that site's 
Consuming address (URL#3), and then, when the user gets to URL#3, if 
authentication was successful they get an instant redirect straight 
back to URL#1 again.

-Shade