Merging Capability Discovery and XRID Proposals into Spec

[David Recordon]

Thanks Mart, I've included TypeKey in the example and updated that bit
of the spec.  Can you also please once over the impact section and make
sure that I got it all right, I'm not quite sure I did.

Johannes, can you look over the notes Mart mentioned please since I
think they are mainly questions aimed at LID?

Thanks,
--David 


-----Original Message-----
From: yadis-bounces at lists.danga.com
[mailto:yadis-bounces at lists.danga.com] On Behalf Of Martin Atkins
Sent: Tuesday, November 08, 2005 12:15 AM
To: yadis at lists.danga.com
Subject: Re: Merging Capability Discovery and XRID Proposals into Spec

David Recordon wrote:
> Right now I'm working on merging Mart's spec changes into the spec on
the Wiki.
> It will be in a little bit of flux as I also work on merging in the 
> XRID proposal as well.  I should be done in the next hour or two and 
> at that point would love to get more feedback especially around how 
> XRIDs have been integrated.
> 

I think my proposal has been integrated just fine from what I can see.
Nice one.

I have a few niggles about the XRID stuff, though:
* The "URI" element should not be mandatory inside "Service". Firstly
both the OpenID and LID capabilities define what happens in the case
where it is ommitted, and secondly not all services will have a URL
endpoint. Consider for example:

<Service>
   <Type version="1.0">http://www.typekey.com/signon<Type>
   <umm:Username>johan</umm:Username>
</Service>

While it is admittedly unlikely that anyone would use TypeKey through
YADIS, it provides a good test case for services that are not URL-based.

Related to this is that there needs to be some more said about how the
extra per-service elements are added. The spec seems to suggest that the
service is free to dream up new XML elements and use them, which might
make some XML-heads squirm. If that was the intention though, it should
probably be made specific in the spec.

The cleaner solution would be to mandate the use of XML namespaces as I
did in my example, but that's an extra overhead for a format that's
currently quite simple and unencumbered by such things. I see that
Drummond Reed's original proposal made use of a namespace for the
"keyInfo" element, but the namespace was not declared.

* Minor nitpick: you left in some of the comments like "I think YADIS
has this" in the format spec. It's probably best that our own spec isn't
so indecisive about what it does! :)