[OpenID] GRDDL and OpenID

[Brian Suda]

On 2/3/07, Sam Ruby <rubys at intertwingly.net> wrote:
> Frankly, he truth is that from a cut and paste point of view, I don't
> think it much matters whether the rel value is 15 characters or 50.  But
> from a spec point of view, html link rel values are meant to be case
> insensitive, something that will pose interoperability issues with URIs...

--- URLs in as REL values would be an abuse of the HTML spec[1].

"[REL] attribute describes the relationship from the current document
to the anchor specified by the href attribute. The value of this
attribute is a space-separated list of link types."

So the value of REL should define the relationship. Values such as
"glossary", "next", "index" make sense because they describe the
relationship. A URL would not serve this purpose.

I think the original thread came from comments on
http://dig.csail.mit.edu/2007/id/talk19 and the issues of using a
pseudo namespace "openid.server"

* Remember the HTML tag wars? table, blink, marquee, etc.
* Are values of rel= all that different from tag names?
* openid.server is less likely to collide than server, just like emacs
lisp function names are less likely to collide if you prefix them with
your initials. Boring
* WebArch 101: Use URIs.

I can't speak for exactly what was meant by "Use URIs", but i would
suspect something related to the following:

Instead of using "openid.server", or "openid.?????" and namespacing
things with the prefix of "openid" use a profile URI to define @rel
and @class values and simply use "server" or "identity-server"

To do this the HTML element has a profile attribute[2]. That profile
attribute descrives the values in @rel, @class. There is no need to
prefix values with "openid" and overload the rel value as namespaced
xml elements values.

-brian

[1] - http://www.w3.org/TR/html401/struct/links.html#h-12.3.1
[2] - http://www.w3.org/TR/html401/struct/global.html#didx-profile

-- 
brian suda
http://suda.co.uk