W3C home > Mailing lists > Public > public-html@w3.org > November 2008

Re: An HTML language specification

From: Mark Baker <distobj@acm.org>
Date: Tue, 18 Nov 2008 12:17:02 -0500
Message-ID: <e9dffd640811180917q1531dcc9tae0dada72132e792@mail.gmail.com>
To: "Geoffrey Sneddon" <foolistbar@googlemail.com>
Cc: "Ian Hickson" <ian@hixie.ch>, public-html@w3.org

On Tue, Nov 18, 2008 at 11:20 AM, Geoffrey Sneddon
<foolistbar@googlemail.com> wrote:
> On 18 Nov 2008, at 15:11, Mark Baker wrote:
>> I skimmed through WF2 forms and certainly saw that a lot of
>> declarative features are defined in terms of the DOM, e.g. "output",
>> but that just seems an editorial convenience - or perhaps more
>> accurately, that as Web & former browser developer, you're just
>> reusing a tool that you're very familiar with.  I see no reason why it
>> couldn't be defined completely independently of the DOM.
> What do we gain by defining it independently from the DOM though? We would
> end up with a longer spec (in all probability) for what I see as little
> gain. There is nothing in HTML5 that requires a DOM to be used,

When the language is defined in terms of a DOM, and some
implementation doesn't have one (e.g. authoring tool, spider) , then
what are its implementers supposed to do?

> but spec
> readers en masse are used to the DOM (as opposed to any other object model
> we could come up with, as we generally need one seeming we're dealing with a
> format with multiple serializations), so defining it in terms of the DOM
> seems to simplify the spec for everyone.

I would say that it's less work for the editor in the short run, and
also probably less work for browser implementers in the short run,
because everything they need is in one convenient location,
consistently edited by one person.

Beyond that, I see no advantages.  After a few years pass and the
protocol definitions need updating, I see only problems for it because
no attempt has been made to decouple the individual protocols: the
dependencies will get only more complex, not less, and the spec will
grow in size and complexity to the point where the cost of
implementation becomes prohibitive... a result which I know is the
exact opposite of what Ian wants to achieve.

Can anybody name a long deployed system whose constituent protocols
were specified as a monolith?  I can't.

Received on Tuesday, 18 November 2008 17:17:43 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 29 October 2015 10:15:39 UTC