[OpenID] Summarizing my grouse with XRD

[Breno de Medeiros]

No, there are no differences. Caching by the Subject is the wrong
thing. If you are starting the discovery from a resource different
from the Subject, then caching by the Subject will not actually save
you network fetches. Anyway, that is not how HTTP caching works,
caching is based on where you got the resource from, not on some clue
embedded in the document.

Basically, for this to work with existing cache infrastructure, cache
should work at the HTTP layer, and validation or trust reasoning be
performed at a higher (XRD-knowledgeable) level.

On Wed, Oct 21, 2009 at 9:07 AM, Santosh Rajan <santrajan at gmail.com> wrote:
> Comparing robots.txt with an XRD is like comparing "apples with oranges".
> Can you do better than that? Cacheing robots.txt is not the same as cacheing
> an XRD. I will explain.
> If my browser wants to cache all my XRD's. This is a real possibility. I may
> have XRD's at Google, Yahoo, Microsoft and "my own" host. The only way you
> can differentiate between all these XRD's is if the XRD;'s have a <Subject>.
>
-- 
--Breno

+1 (650) 214-1007 desk
+1 (408) 212-0135 (Grand Central)
MTV-41-3 : 383-A
PST (GMT-8) / PDT(GMT-7)