Proposal for an XRI (i-name) profile for OpenID

Martin Atkins mart at degeneration.co.uk
Tue Apr 4 07:12:36 UTC 2006


Drummond Reed wrote:
> Martin's right here except one detail -- when the Relying Party resolves the
> XRI using standard or proxy XRI resolution, it will get back the Yadis XRDS
> document. No need to "resolve it to URL, then do Yadis discovery on it". The
> standard response to XRI resolution *is* an XRDS document.

This does raise the question of what URL is used to do the OpenID bit,
though.

If my identity URL is http://frank.livejournal.com/, but that returns an
HTML document that points to http://frank.livejournal.com/data/yadis,
what do I point my XRID at?

LiveJournal's OpenID server doesn't know how to deal with the
/data/yadis URL, because by design the OpenID server has no knowledge of
Yadis. The OpenID server needs to "see" the canonical identity URL.

So if I've got an XRID that resolves to that /data/yadis URL, I *must*
do openid:delegate to http://frank.livejournal.com/ or it won't work.
Note that LiveJournal's Yadis output doesn't currently do that.

It seems to me that Yadis' indirection-through-HTML trick and XRID
resolution are incompatible with one another. Am I missing something?

(Is it even possible for someone to set up an XRID that points at a
LiveJournal XRDS document without modifying LiveJournal's XRDS document?)



More information about the yadis mailing list