[OpenID] Thoughts on many openids to one account and xrds

[Mark Atwood]

I just got a Dopplr account, and it's somewhat OpenID enabled.  One
neat thing it does is that one can apparently associate several
different OpenIDs with one's account.

When I did that, I gave it my preferred OpenID, http://mark.atwood.name/
and it proceeded to use the link rel openid.* links, etc etc
as expected.

However, I also have a line
<meta http-equiv="X-XRDS-Location" content="http://mark.atwood.name/me.xrds" />
in that index.html file, and said XRDS file lists a bunch of openids
that are mine.

Do the existing libraries and implementations make it doable
and easy for sites to discover and parse XRDS data?

I think it is wise to encourage sites to do so, and associate all the
OpenIDs so discovered with that user account.  And perhapse also to do
so recursively, as each OpenId is checked, still more XRDS data can be
discovered.  And a best practice would be to periodically do
appropriate HTTP HEAD calls, check for modifications, and keep up to date.


This is not a criticism of Dopplr in any way, just a general
suggestion as to something we should encourage and make easy for sites
to do.

-- 
Mark Atwood                 When you do things right, people won't be sure
me at mark.atwood.name         you've done anything at all.
http://mark.atwood.name/   http://fallenpegasus.livejournal.com/