Re: W3C/WHATWG overlap going forward

On 28/11/2014 09:34 , Henri Sivonen wrote:
> On Wed, Nov 12, 2014 at 5:28 AM, Sam Ruby <rubys@intertwingly.net> wrote:
>> 1) A number of individuals within the WHATWG would like to see the W3C no
>> longer copy and/or work on areas that overlap with ongoing WHATWG efforts.
>
> I think that *if* the W3C is going to specify things that do overlap
> WHATWG efforts, rewriting things to "not copy" is worse than copying,
> since rewrites are likely to introduce subtle deviations. I think the
> ideal outcome would be for the W3C normatively reference WHATWG specs
> such that the normative statement get imported by reference under the
> PP; see the last sentence of this email.

I agree that having happy references between W3C specifications and 
WHATWG specifications (and any other specifications) would be great. But 
there's a granularity problem.

In order to avoid talking about anything real that might hurt anyone's 
real feelings, I'm going to go with a completely fictitious example.

Let us imagine that, working within the confines of the Shiny Donkey 
Coalition I wish to produce a specification for a magnifier that can 
focus sunlight in just the right way to grill toast. It so happens that 
the fine people at the Rainbow Unicorn Covenant have an absolutely 
excellent description of the algorithm involved in grilling toast, such 
that it comes out perfectly browned and warm. Unfortunately, it is part 
of the Rainbow Unicorn Kitchen Specification that also contains 
completely inane remarks about how one should butter one's toast, as 
well as some wishful thinking about fridges that can automatically order 
pizza.

I think we're all agreed that publishing a specification for Slightly 
Different Kitchens isn't ideal. The open question is: what is?

I think that there are several options, that aren't mutually exclusive 
nor necessarily exhaustive.

Perhaps someone — Shiny Donkey, Rainbow Unicorn, or from the Dahut 
Liberation Front (it doesn't matter) — can produce a better Toast 
Buttering Best Practices and, after some time passes and it clearly 
emerges as the best way to butter your toast, it's what everyone 
references. I don't think that this action should be prefixed with a 
long discussion — it is well-known that long discussions about butter 
easily go rancid.

Perhaps the Shiny Donkeys can develop a smart but easily comprehensible 
scheme to reference just parts of a given specification. (Bikeshed 
definition autolinking comes to mind.)

Perhaps the Dahuts establish a better way of toasting such that the 
bread stays squishy in the middle and, within the context of 
SunToasting, the problem goes away.

None of these is perfect, there can be transitivity problems (though I 
think that in implementation practice those will largely be red 
herrings), but we can work through solutions. Sure enough it would help 
if WHATWG HTML were easier to work with for people who wish to build new 
features, but absent that we'll collectively hack our way there.

>> 3) We discussed proposals for modularity
>
> This seems to contradict the point quoted above about limiting the
> differences between WHATWG HTML and W3C HTML.

To put things in perhaps too small a nutshell, limiting gratuitous 
differences should not come at the cost of abandoning useful ones. And 
useful ones are best realised in their own module.

>> 3) In 1Q15, we will identify which specs in the WHATWG the HTML WG would
>> like to refer to directly, and will schedule a call with the director to
>> build a plan.  Examples of documents that could be referenced by W3C HTML
>> Work: Fetch, URL, Streams.
>
> Cool. If Fetch, URL and Streams become referencable by the W3C, why
> not WHATWG HTML itself so that the W3C could focus on extension specs?

Do you think it would be possible to address the granularity issue from 
the WHATWG end?

> I wonder if this WG could, from time to time, publish a REC that, in
> addition to boilerplate, just says "This Recommendation normatively
> references the snapshot of WHATWG HTML dated yyyy-mm-dd as if its
> normative statements were included herein." (Where yyyy-mm-dd is
> enough in the past for Members to have time to be comfortable. The
> current track record suggests this doesn't need to be more than a year
> in the past. I suppose continuing the statement with "except for X, Y
> and Z" would still be better than the W3C publishing a different
> document or, worse, a set of modularized documents and leave it as an
> exercise to the reader to figure out the diffs.)

I don't know if it would be under the specific form that you are 
suggesting, but at some point I would love it if we could publish a 
specification that basically says "The Web Platform is this, that, and 
the other thing over there." But we're a number of collective steps away 
from that.

-- 
Robin Berjon - http://berjon.com/ - @robinberjon

Received on Wednesday, 10 December 2014 16:35:45 UTC