Anne van Kesteren

Standardize, don't try to invent

After reading the The Atom End-Game I now feel even stronger that the WHATWG is heading the right direction.

I’ve been involved in several different standardization projects across the years, of which one was overwhelmingly successful: XML. And in the process of designing XML, we invented more or less nothing. We took an existing standard, SGML, parts of which worked well and other parts of which were klunky, or expensive, or incomprehensible, or all three. We threw away everything but the pieces that were known to work and added pretty-good Unicode support, i.e. something else that had been proven to work. We tightened up some definitions and added some convenience features and threw away lots and lots and lots of options.

Standard organisations should not try to invent new specifications, they should look at existing implementations of a certain technology and try to standardize what works.

Comments

  1. Who then innovates/invents new things?

    Not a bad idea though, but it's a hard sell.

    Posted by Jeff Minard at

  2. Somebody should have said this years ago to the current XHTML2 working group...

    Posted by Robbert Broersma at

  3. Anne blogged:

    Standard organisations should not try to invent new specifications, they should look at existing implementations of a certain technology and try to standardize what works."

    The W3C tried that with HTML 3.2, which set HTML authoring back a decade. Its only thanks to web standards evangelists have we managed to claw our way back to structured and appropriate markup.

    We've only just reached the quality of markup documented in the HTML 3.0 recommendation.

    Posted by Isofarro at

  4. Isofarro, we're not even there yet. But you're absolutely right. Some standards need to specify existing, working practice, while others need to do something new. We need both.

    Posted by Asbjørn Ulsberg at

  5. Here's an interesting innocent victim of war rant I got today.

    Posted by Randy Charles Morin at

  6. Anne, did you know the Preview accepts this notation,
    <a href='asdf'>
    but it's rejected on the Post?

    Posted by Randy Charles Morin at

  7. Asbjørn - while you're right that there should be specifications for newly-invented things, I think Anne's right that standards bodies shouldn't be the ones inventing those things. Standards bodies should stick to standardizing things already in use, not specifying new things.

    P.S.: Randy, that's because of the single quotes. I've had problems with it in the past. (Can't remember if I ever bothered to tell Anne or not...)

    Posted by dolphinling at

  8. Thanks dolphinling

    Posted by Randy Charles Morin at