RE: On ISSUE-41: Decentralized extensibility

Smylers wrote:
>Chris Wilson writes:
>> I think extensibility is a very good idea,
>
>Hi there.  To help with understanding this, could you give a few
>examples of the kind of extensions you're thinking about here?

Primarily extensibility in order to make it easier to incorporate future vocabularies not automatically added to the compendium of HTML5/HTML6/etc. - e.g. when HTML4.01 was standardized, we weren't thinking about including SVG into it.

>I'd've thought that any future vocabulary will require certain behaviour
>to go with it, such that a standard will need to specify the behaviour
>and browsers will need to implement it.  That sounds like a considerable
>amount of work.  Would working out the syntax now, rather than then,
>significantly ease the burden of adding this vocabulary?

Yes.  We wouldn't have the current discussion about including SVG each time, for example.  It might also be easier to think about fallback generically - or, for example, to point to more information about the vocabulary.

>Or is the benefit that browser engines could generically parse all
>extension vocabularies (without knowing their semantics) and hand off
>implementation to plug-ins, such that no core-browser changes are needed
>for such new vocabularies?

In some instances, yes - though this is difficult to do well, but I can certainly imagine once we have an interoperable binding language, for example, it might be easier.

>Are you expecting many such new vocabularies?

I expect there will be more than I expect.  :)

>> (or we're just encouraging people to roll their own).
>
>Interpreting "people" as "the group writing the standard defining what
>the vocabulary does", why would that be a bad thing?  They are
>specifying everything else to do with it, so are probably in a good
>position to specify the syntax as well.

Not really, because it has to work inside an HTML engine.  Or they can just do XML and we can sideline ourselves.

>If we specify the syntax now (without knowing what these hypothetical
>new vocabularies are) then surely there's a risk of us not doing it as
>well as the group who are experts in the vocabulary in question?  We
>could end up putting inappropriate restrictions on them, such that their
>vocabulary is hampered into a less convenient form, or that they ignore
>our syntax entirely (making it pointless).

Not really, IMO - if they can't live with the restrictions, then they can have their own syntax that cannot embed in HTML.  It's a tradeoff.

>> I think it's a poor language that doesn't think about its own
>> extensibility,
>
>XML obviously does extensibility.  All HTML features are available in an
>XML language, XHTML.  Why do we need for HTML to be arbitrarily
>extensible with any vocabulary, when we already have XHTML which does
>that?

Because I don't think we really have XHTML.  Or maybe I'm misreading interest in strict XML implementations of HTML5.

-Chris

Received on Wednesday, 23 July 2008 23:20:46 UTC