On 2/1/07, <b class="gmail_sendername">Brian Suda</b> <<a href="mailto:brian.suda@gmail.com">brian.suda@gmail.com</a>> wrote:<div><span class="gmail_quote"></span>> There was a small talk about OpenID and GRDDL:<br>
> <a href="http://dig.csail.mit.edu/2007/id/talk19">http://dig.csail.mit.edu/2007/id/talk19</a><br><br></div>In the presentation referenced above, Dan Connolly of W3C repeatedly makes the point that he believes that URI's should be used instead of string values as the value of "rel=" attributes. One might argue that a string is a string, thus, other than the fact that URI's are typically longer than non-URI strings like "
openid.server", it shouldn't make much difference to any OpenID programmers if the "rel=" values are strings or URIs. Thus, if URIs make some folk happy and if their use can enable more easy integration with other systems or ease various forms of inference, it might make sense to simply make the switch and use URI's in future drafts.
<br><br>Are there other arguments in favor of using URIs rather than strings? <br>Has this been debated on the list before? (sorry if I've missed something...)<br>What negative results would come from redrafting OpenID
V2.0 to use URIs as "rel=" values?<br><br>bob wyman<br><br>