W3C home > Mailing lists > Public > public-html@w3.org > February 2009

RE: ISSUE-4: Versioning, namespace URIs and MIME types

From: Ian Hickson <ian@hixie.ch>
Date: Thu, 19 Feb 2009 00:08:09 +0000 (UTC)
To: Larry Masinter <masinter@adobe.com>, Philip TAYLOR <Philip-and-LeKhanh@Royal-Tunbridge-Wells.Org>
Cc: Maciej Stachowiak <mjs@apple.com>, HTML WG <public-html@w3.org>, Dan Connolly <connolly@w3.org>
Message-ID: <Pine.LNX.4.62.0902182357390.6186@hixie.dreamhostps.com>
On Wed, 18 Feb 2009, Larry Masinter wrote:
> >>
> >> Implementations that support more than one language or incompatible 
> >> version need to, along with the code that can generate or access a 
> >> DOM, remember the language or version associated with the code.
> > 
> > That would violate the spirit of our DOM Consistency design principle.
> The spirit of principles need to be examined against the realities of 
> compatibility and practice.

In what sense has it not been so examined?

> The reality of practice seems to be that there are "versions", currently 
> switched on by DOCTYPE, for "quirk mode" or "standard mode" and that 
> some browsers switch between one version and another depending on the 

There are in fact four versions in most browsers, with a fifth mode in 
IE8. (HTML5 calls the first four modes "quirks", "limited-quirks", 
"no-quirks", and "XML document"; IE8 calls the fifth mode "IE8 mode".)

I have never heard any browser vendor talk about any of these modes in a 
positive light, only ever as a necessary evil, and often with regret. 
Adding mode modes should be on our list of things to avoid wherever 

> The issue, ISSUE 4, is the about versioning mechanism for HTML5. ISSUE 4 
> is in scope for this mailing list. 
> http://www.w3.org/html/wg/tracker/issues/4
> I accepted ACTION-93, to make a proposal on doctype and versioning. 
> Discussing ACTION-93 is in scope for this mailing list. 
> http://www.w3.org/html/wg/tracker/actions/93

Then my original e-mail on this thread is relevant in that it shows that 
we only have two possible axes for versioning, namely the namespace and 
the tag name for whatever elements we want to version.

Versioning in general is bad idea IMHO, as discussed almost two years ago:


That e-mail also links to detailed rebuttals of versioning by various 
other people.

On Wed, 18 Feb 2009, Philip TAYLOR wrote:
> Maciej Stachowiak wrote:
> > 
> > So I think experience since David's original email bears him out. It 
> > does seem to make a difference whether or not the spec has an official 
> > versioning hook. It doesn't prevent versioning mechanisms for bug 
> > compatibility from being created, but it does limit their 
> > effectiveness in producing anti-competitive side effects.
> Without wishing to over-simplify things, is this concern about 
> "anti-competitive side effects" what lies at the heart of much of the 
> opposition to versioning ?  If so, it is a pity that this wasn't made 
> explicit at the outset.

I first referred to it explicitly as such on this list in April 2007:


David said as much even earlier:


Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'
Received on Thursday, 19 February 2009 00:08:44 UTC

This archive was generated by hypermail 2.4.0 : Saturday, 9 October 2021 18:44:43 UTC