Re: An HTML language specification

On Mon, Nov 17, 2008 at 2:28 AM, Ian Hickson <ian@hixie.ch> wrote:
> On Mon, 17 Nov 2008, Mark Baker wrote:
>> On Fri, Nov 14, 2008 at 4:23 PM, Ian Hickson <ian@hixie.ch> wrote:
>> > Could _you_ elaborate in detail on what makes a "language" spec vs
>> > whatever the HTML5 spec is?
>>
>> A language specification defines the language
>
> This seems tautological. Could you elaborate?

For a markup language, it defines the meaning of the elements and
attributes, and their composition.  e.g. it defines what "img" means,
and how "alt" affects that meaning.

>> and the mappings to and from a serialized form.
>
> I agree with this part, although it seems not everyone else does.
>
>
>> In other words then, I believe HTML 5 should be limited to (roughly)
>> sections 1, 3, and 8 of the latest published WD.
>
> You would exclude section 2 (in particular 2.4), which defines the
> syntaxes allowed in all the attributes,and section 4, which defines the
> vocabulary? I would have thought section 4 was pretty much the most
> important part of any "language" specification.

8-)  I was referring to the latest published WD.  I see you've
renumbered the sections in the editor's draft.

Yes, I would include all of that.

> Presumably you intend this set of sections to indicate a desire to exclude
> the DOM APIs from the language, but wouldn't that mean that SVG was also
> not a language specification? What makes SVG a language specification when
> HTML5 is not?

Yes, I forgot SVG included a DOM, so shouldn't have mentioned it.

And just to be clear, both the current HTML 5 and SVG specs do include
a specification for a language.  They just also include a lot of other
stuff which isn't required for a language specification.

> I should note that excluding the DOM APIs from HTML5 is really not an
> option. HTML4 and DOM2 HTML tried that and we ended up with a disaster:
> the APIs and the vocabulary aren't defined together, so features like
> <video> are effectively impossible to introduce; and the interaction of
> the declarative features, the DOM features, and the dynamic behavior (e.g.
> the behavior of navigating a browsing context) become incredibly hard to
> define if all three parts are in different documents.

I can certainly see ample complexity in the interaction of DOM and
scripts/objects, but can't see how the declarative aspects would enter
into that complexity.

> One of the biggest goals of HTML5 is to fix up the mess that was left from
> leaving things like Window and the navigation algorithm undefined. These
> are core to how HTML works. We can't leave those out.

IMO, those are not core to how HTML works, they are core to how HTML
with scripts is processed by browsers, which is a very different
thing.

Mark.

Received on Monday, 17 November 2008 18:06:42 UTC