Re: review of content type rules by IETF/HTTP community

On Mon, 2007-08-20 at 16:40 +0200, Julian Reschke wrote:
> Dan Connolly wrote:
> > Indeed, early review feedback[Fielding07] suggests I
> > should cite/excerpt more of the background that I had in mind
> > before sending this out for wider review.
> > ...
> 
> I think it's clear that this is a point where theory and current 
> practice do not match. Also, the situation might be much better if at 
> least some of the browsers would allow the user to opt-out of content 
> sniffing, as RoyF suggested multiple times.
> 
> Documenting what User Agents do today with respect to sniffing is 
> certainly a good idea. What's not clear to me is why this needs to be 
> part of the HTML5 activity. If the browser vendors want a common spec 
> summarizing what their products need to do today, fine. But why does 
> this have to be part of HTML5?

It's currently part of the HTML 5 work because it's in the
text that we adopted for review on 9 May and we haven't taken it out.
http://www.w3.org/2002/02/mid/46423D1F.5060500@w3.org;list=public-html

According to my understanding of Web Architecture, it's quite
a wart on the HTML spec; it belongs elsewhere.

But the principle of Well-defined Behavior argues for including it.
http://dev.w3.org/html5/html-design-principles/Overview.html#well-defined-behavior

I find that principle unappealing, but I'm somewhat persuaded
that it's a necessary evil, or at least that it's cost-effective.

And of course, I consider the current content-type handling in
the browsers a bug and I'd like to see it fixed. But it does
seem like a browser that does so would lose market share
(see also "Support Existing Content"), so I'm not holding my breath.


-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Monday, 20 August 2007 15:09:29 UTC