[OpenID] host-meta and "acct:"

[John Kemp]

On Nov 5, 2009, at 11:13 PM, SitG Admin wrote:

> I think I have a better understanding now. I will try to explain it:
>
>> Are you suggesting that you guys know something which those guys  
>> didn't know?
>
> No. But . . . forward-compatibility.
>
> History has shown us that we (and others) will continue to have  
> ideas that are not covered by existing specs, requiring us to either  
> come up with new specs or modify existing specs. We don't need to  
> specifically anticipate these ideas, we don't need to know all the  
> details, to foresee that they (probably) *will* be pursued. If, at  
> that time, the XRD spec does not support them, their development  
> will be slowed down, and their implementors may go make a rival spec  
> to accomplish/support what *they* are after, faster than XRD can be  
> modified to allow for it. This splinters what would otherwise be a  
> single XRD community into several similar groups, and divides the  
> attention of those whose interest is rooted more in the compatible  
> ideas than steadfast loyalty to particular specs.

Yes, but the question is what does XRD actually stand for? What  
semantics does it have? It's great to be extensible, but history has  
shown it is better to make hints about what kinds of extensibility are  
interesting rather than simply give a blanket right to extend.

>
> By putting greater flexibility into the XRD spec than any specific  
> *need* has been shown for, compatibility with future ideas is made  
> more likely. Today the Subject may be necessary for 99% of use  
> cases; that percentage may change again in future, though.

 From the use-cases I've heard on this list so far, there is ALWAYS,  
conceptually at least, a Subject - whether it is a "host", or an  
explicitly URI-specified Subject, or a Subject which is the "user  
account that I have just done discovery on", there is always,  
conceptually, a Subject.

And, from earlier discussion, I heard the idea that without a Subject,  
XRD becomes a template spec. which someone else has to "complete" with  
their own use-case. Would it not then be worthwhile to create a  
Subject XML element that could carry information which could be used  
to capture all notions of a Subject without saying "Subject is  
optional" when in fact it it seems NOT to be so, logically or  
conceptually-speaking? How important is the CONCEPT of a Subject in  
XRD? I think that it is very important, and in fact, likely not  
optional at all, at the conceptual level.

Regards,

- johnk