A post by David Hyatt about web standards reminded me of this rant that's been circling in my head for a while.

Last year, I ranted on mozilla.general about web standards and the W3C.

I mean, what's with the W3C? CSS2 was released in 1998 as a standard, and I can count on the toes of one hand how many compliant browsers have been released, three years later. So what are the W3C doing about it? Yes, they're off working on CSS3. Their own HTML showcase, Amaya, doesn't even support CSS1 properly.

Wahey!

The closest we have to an implementation of this three year old standard is Mozilla. So it's not really "W3C CSS2", it's Mozilla CSS2, because nobody else has bothered with more than a third of it. It may be W3C DOM, but it's an incompatible subset of what people are really _using_ out there.

A follow-up replied to mention that Mozilla is closely involved with the W3C, and you could consider it to be the reference implementation if you want.

However, my sentiment stands. A standard should not be given the name "standard" until there exists a real-world implementation. Until something concrete exists that demonstrates the standard, it's just a good idea that might some day work.

Without an implementation, a standard is unproven. There may be significant, vague edge-cases. There may be outright self-contradictions. There may be things missing that people really want to be able to do (text flowing between columns, anyone?). There may be things that nobody really wants to do, but that catering for adds 10% to the running time. Until there's an implementation, nobody knows.

Publish a standard without an implementation, and you leave those vague edge-cases and self-contradictions to be worked out by the implementors. You'll have competing implementations of the same standard that don't behave the same way. Have a reference implementation, and you can say "It has to look like this"

Anyway, enough ranting for now.