YADIS as an abstraction layer

Martin Atkins mart at degeneration.co.uk
Tue Nov 1 14:22:19 PST 2005

Granqvist, Hans wrote:
>>Most human-readable things on the web are HTML. People are familiar with
>>HTML. There are lots of people that can write HTML but don't even know
>>what HTTP headers are, let alone how to change them.
>>. . .
>>Sure, it's not ideal from a technological perspective, but
>>pie-in-the-sky pure implementations that don't pay any mind to current
>>realities rarely get very far.
> Oh, I agree with this. Most of it.
> The reliance on HTML should be made explicit somehow . . .  People don't 
> break protocols. Implicit dependencies break protocols. ;)
> Perhaps: the requested format of the response is part of the request, a la 
> "format=xyz" param/value, where HTML is the default value, or similar.

The consumer doesn't want HTML, it wants application/x-meta-identity (or
whatever we end up using.)

It gets HTML because the site owner can't or won't configure things so
that YADIS consumers see one thing (a meta-identity document) while
everyone else sees another (an HTML document).

HTML is not a dependency as such, we just allow for it as a special case
because that's the way things are right now. If HTML falls out of
favour, you can always just serve a meta-identity document directly.

The only awkward part is that all consumers must include some form of
HTML parser. This was made easier in OpenID by artificially restricting
the syntax of the LINK element (while still remaining valid HTML/XHTML)
so that it could be "parsed" by regex rather than a full-on HTML parser.
It's a bit icky, but it's all just protocol.

More information about the yadis mailing list