Re: <canvas> and the 2D context API (was RE: Begin discussions for pushing Last Call into 2010)

Maciej Stachowiak wrote:
>
> On Aug 12, 2009, at 5:56 PM, Shelley Powers wrote:
>
>>
>> If we were to split the API into a separate document, could it 
>> progress at a different pace than the main HTML 5 specification?
>
> To a limited extent, yes. But HTML5 would likely have a normative 
> reference to a Canvas API spec, so there is a limit. Per W3C rules, 
> you can't normatively reference a document that's more than one 
> maturity level apart. So we couldn't go to CR (and thus couldn't enter 
> LC) until the Canvas API draft reached Working Draft status, and we 
> couldn't go to PR until the Canvas API draft reached CR.
>
>> Could it also have a different editor or set of editors?
>
> Yes. But I expect Ian is unlikely to remove the 
> current CanvasRenderingContext2D section until someone demonstrates 
> they can do the work by actually producing a viable separate draft. 
> And I expect much of the Working Group would feel the same way.
>
>> (Note though that even if we take out the 2D graphics context, the 
>> element still belongs in the HTML spec, as it's part of the language. 
>> So technically "<canvas>" still would be in the spec; just the 
>> graphics context API would be taken out. One could argue that that 
>> would lead to the spec being overly confusing to implementors, who 
>> generally prefer things in one place to implement them, as it leads 
>> to fewer "cracks between the specs".)"
>
> I do think there is the risk of confusion in such a case.
>
>> Now, I do agree that the Canvas _element_ belongs in the HTML 
>> specification, but the API should be split out. Doing so would also 
>> work in well with the ongoing effort to incorporate accessibility.
>
> To the contrary, I think splitting out Canvas could harm the 
> accessibility work, for reasons stated in my other email.
>
>
> Generally speaking: Over the past two years, a number of people have 
> expressed interest in the existence of a separate spec for the API, 
> but no one has actually done it. The time window to produce a draft 
> that could receive adequate consideration before Last Call is limited. 
> I believe a quality draft would receive fair consideration. Note that 
> this is considerably more complicated than just cut/pasting a big 
> chunk of text out of HTML5. That would leave many dangling references 
> and unclear definitions in both directions. But I believe that it is 
> doable, and there would be some technical advantages to making 
> the CanvasRenderingContext2D API more readily reusable. I also suspect 
> Ian would consider doing the work himself post-HTML5, if no one else 
> gets to it by then.
>
> Personally, I don't think it's worth raising this as an issue until 
> someone produces a draft or volunteers to do so on a reasonable 
> deadline. This task has not really gone anywhere as a wishlist item.
>
> Regards,
> Maciej
>
I don't see any usefulness to splitting out one section of HTML 5 into 
other HTML 5 document, being bound by the same rules, editor, and 
timeline. That defeats the purpose. All that does is make the main 
document smaller, but it still ties Canvas into being constrained by the 
HTML timelines.

I believe it is doable to split the section out, and yes clean up 
dangling references, into a separate specification, but I'm not sure 
there's enough time to get it accepted as a W3C working group and prelim 
specification AND somehow pull it out of HTML 5.  The fight for the 
latter is most likely what turned most people off.

This is unfortunate, because it has no place in a specification on HTML, 
nor is it good for the Canvas element, as it will become uncompetitive 
very quickly.

I'll wait to hear what Adrian says, but I don't see how anything can 
happen in a month's time.

Shelley

Received on Thursday, 13 August 2009 12:21:58 UTC