[whatwg] Versioning

Robert Brodrecht whatwg at robertdot.org
Wed Mar 14 11:27:21 PDT 2007

Anne van Kesteren Wrote
> If for Internet Explorer having
> a   real standards mode is unavoidable for some reason I'd suggest (and
> have)   that they use <!doctype html> in text/html and XML in general to
> trigger   it.

I'd love to use their super-standards mode in HTML4 and / or XHTML.  Or
maybe use it some time in the next few years, if it were available before
HTML 5 was completed.  Using the HTML 5 doctype will not allow me to do

> I'm not sure what this has to do with the other points made in this
> thread.

I was replying to your assertion that authors "code against
implementations, not specifications" by:
1) Giving a good reason why not to code against implementation only
(because the author can make up arbitrary elements and use them in
practice, hence the examples).  Creating proprietary markup does not
ensure the longevity of the document, breaking with the ideals of
standards.  Coding against implementations allows me to create proprietary
2) Say that validation is a big part of the web standards movement. 
Validation is coding against specifications.  Therefore, at least in the
standards movement, people code against implementations AND
specifications, and at least attempt to create valid markup, even if they
don't always succeed.  Not just implementations as you asserted.
3) Because, ultimately, a policy of letting implementors dictate standards
as-they-go doesn't seem like a good idea.  (However, you've cleared that
up some, and I hope you mean something more along the lines of: if the
innerHTML functionality in JavaScript implementations is added to the DOM
specification; in that case, the implementation is providing much needed
functionality that the specification should have had in the first place. 
That is different than, say, Microsoft changing the specs for the W3C box
model because IE's bad implementation had a large user base.)

That said, this has ended up completely off topic.

Robert <http://robertdot.org>

More information about the whatwg mailing list