XMLK: A blast from the past

Volume 8, Issue 162; 16 Dec 2005; last modified 08 Oct 2010

Back in 2003, there was a flurry of discussion about “simplifying” XML. Last night, Tim asked me about a proposal that I'd made. It took me a while, but I finally found it.

Back in 2002 and 2003, there was a flurry of discussion about “simplifying” XML. It fell to the TAG under the rubric xmlProfiles-29. As I recall, some of us were worried about a loss of interoperability if specifications began to follow the lead of SOAP and accepted only a subset of XML. SOAP doesn't allow a doctypedecl, which is clearly part of the grammar of XML, and so it's possible to argue that SOAP doesn't use XML. I'm sure I've made that argument on occasion.

One possibility we explored to address this concern was to define a subset of XML and give it a name. This, we hoped, would allow everyone who wanted to subset XML to agree on the same subset and thus preserve interoperability.

I forget exactly what motivated me, but one afternoon I set out to write the smallest subset specification I could. I posted the result to the W3C member-only TAG mailing list, and we discussed it. I even got an action to publish my specification on the public TAG mailing list.

Fast forward almost three years and nothing has really changed. An official subset has never been developed, but neither have the interoperability problems we feared.

Anyway, last night Tim asked me for a pointer to my minimal specification. My local copy was easy to find, but why couldn't Google find it on the web? Because apparently I never completed my action item! I described it on xml-dev, so it did wind up in a public space, but not as a specification.

So here, for posterity, is the XMLK specification I wrote. Now we just have to wait and see what Tim's writing about. :-)


Quick note, the relative link to the XMLK specification works from the Web page for the post, but not when reading the post from a feed. An absolute link would be marginally better. Cheers, Tony.

—Posted by Anthony B. Coates on 16 Dec 2005 @ 09:07 UTC #

Absolute links are a pain in the authoring environment. Instead, I added an xml:base attribute to the atom:content element in the full text feed. If you can talk your reader into reloading it, that should fix the problem. I think.

—Posted by Norman Walsh on 16 Dec 2005 @ 09:34 UTC #

Now the only thing that is left is to have a specification that combines this with a requirement on namespaces and get specifications to use that. Validation is overrated.

—Posted by Anne van Kesteren on 16 Dec 2005 @ 10:53 UTC #

By the way, somone pointed out to me that your draft makes the XML Declaration optional. I assume that was not the intend as it builds on top of XML 1.1?

—Posted by Anne van Kesteren on 16 Dec 2005 @ 11:30 UTC #

Hi, Norman,

how does your browser work with the following URL of yours?


(somehow "h" is missing from http).


—Posted by Matej Cepl on 06 Jan 2006 @ 05:42 UTC #

Sorry about that Matej. It was just a typo. Fixed now.

—Posted by Norman Walsh on 06 Jan 2006 @ 07:42 UTC #

Anne, with respect to the XML Declaration and 1.0 or 1.1, I don't think I feel strongly about it. Back in '03, I was probably more optimistic about the near-term future of XML 1.1. If XMLK went anywhere, I wouldn't mind if it was built on top of 1.0 and 1.1. Not that I think it's ever going to go anywhere.

—Posted by Norman Walsh on 06 Jan 2006 @ 07:44 UTC #