Anne van Kesteren

XML: Browser edition

I’d like a version of XML aimed at browsers. Or at least some base format on which browsers can implement SVG, XHTML and other XML based formats:

Crazy, but that’s about it. For now, this is probably as much a pipe dream as the semantic web.


  1. Entities only make sense if you have one for every unicode character.

    Posted by Sjoerd Visscher at

  2. Yes, I was thinking that was an oddball request myself to add lots of them.

    Posted by Robert Wellock at

  3. Semantic web isn't entirely a pipe dream. The end goal of what is known as the Semantic Web initiative may not be attainable if at all desirable. The web does at present conform to some semantics, even Microformats are doing there bit. (That Gates plug certainly helps)

    The problem with an idea like Semantic Web and browser XML (which I like btw.) is that in the real world websites are still made by humans, in fact this may never change. Just look at WYSIWYG editors like Dreamweaver. After at least a decade they still don't work cleanly, getting them to generate XML at a stylists whim will prove a bit of a challange.

    A web built by humans means a web that needs to be fault tolerant and thus we will need HTML. For software generated output you could go the browser XML root. A bit like that wizz bang XHTML of the 90's, oh wait, that didn't really work either, did it? :(

    Posted by Egor Kloos at

  4. What benefit is there in combining XML and XMLNS into one spec? Unless you just meant the new XML spec should normatively reference XMLNS and make it a conformance requirement.

    I agree with dropping dropping DTDs, since they're useless on the web anyway. If one wants to use DTDs for validation, they can still do so but there's no need to include them in the published document.

    I don't agree with extending the list of predefined entities, I think such things should be stripped out either before or during publishing and replaced with the characters they represent (or numeric character references is the encoding used doesn't support them). There's nothing wrong with an authoring tool supporting entities for that purpose, but, like DTDs and DOCTYPEs, there's no point for them being in the publishd document.

    Posted by Lachlan Hunt at

  5. XML without XMLNS is near useless, no? There's no way to include semantics without a namespace. And XMLNS isn't all that useful as a standalone spec, seeing how it's completely dependent on XML. Is there any reason not to combine them?

    And I second everyone else in saying I don't think the predefined entity list should be expanded. The five there are there for a very good reason: escaping. For everything else, Unicode is a better solution.

    As for DTDs... AFAIK, they have at least one good use, which is non-predefined entities. I'm pretty sure e.g. Firefox uses them in XML-based UI for things like version strings, and theoretically a web-based app could do the same.

    Posted by dolphinling at

  6. What about non-user-aggravating-error-handeling ?

    Or is that just up to xHTML?

    Posted by James / AkaXakA at

  7. Yeah, we should come up with a new XML based on the parsing algorithm in HTML5. Many UAs have some sort of non-drocanian error handling for XML — it would be good to have them interoperable. Perhaps HTML5 and this XML can be interchangable, then we can add declarative namespaces to HTML so that you can have "XML islands" in HTML à la IE.

    Or perhaps not.

    Posted by zcorpan at