Re: Formally Object to Referencing WhatWG within the W3C HTML5 specification

On 10 Jun 2010, at 07:11, T.J. Crowder wrote:

> Hi Bijan,
> On 9 June 2010 17:01, Bijan Parsia <> wrote:
> [snip]
>> On 9 Jun 2010, at 15:57, T.J. Crowder wrote:
>> [snip]
>> What I wrote was:
>>> FWIW, completely agree that there must be one specification for  
>>> HTML5. Unless the W3C is prepared to step back and let the WhatWG  
>>> take ownership, that spec must be "owned" by the W3C. Pages like  
>>> this one [ 
>>> ] are very confusing. I've seen it cited in online discussions as  
>>> "the HTML5 standard" (and why shouldn't someone think it was? It  
>>> says "draft standard" on it).
>>> The work of the WhatWG is extremely important, it has driven and  
>>> continues to drive this process forward where HTML had been under-  
>>> and mis-specified for years. That work needs to be credited and  
>>> honored, but as HTML5 is becoming the new baseline, there needs to  
>>> be a single definitive source of normative information about it,  
>>> with other sources of draft proposals (not standards, not  
>>> specifications)
>> You don't mean "not specifications" do you? It's hard to see how  
>> you can propose something without providing a specification of what  
>> you propose.
> You need to specify your proposal, sure. I think there's a pretty  
> clear distinction between doing that and having an alternate HTML5  
> specification, which is what seems to exist at the address I gave.

At one point the working ground *encouraged* the development of  
alternative specifications in the heavyweight sense.

I've seen people think that:
was an alternative specification (and, indeed, should be *the*  
specification). (Now it labels itself as something else, of course.)

This still exists on the web:

The home page of the group lists the main spec as a "draft  

I've seen confusions about various editor's drafts, w3c note, etc. I'm  
not convinced that there's an in principle special thing going on here.

Until Rec, *all* drafts are drafts and thus provisional. Working  
drafts have no special status, explicitly, except for the LC ones (and  
the first, as it triggers aspects of the patent policy).

>>>> very, very clearly labelled as such.
>>> Having a competing "specification" is a sure route to fracture and  
>>> failure. I hope no one wants that. Those of us relying on these  
>>> standards certainly don't.
>> There are two HTML 4 standards, the W3C "recommendation" and ISO/ 
>> IEC 15445:2000(E):
>> Some purists might insist that only the latter is a standard :),  
>> but there are definitely two specifications. Now, editorially, the  
>> ISO spec is a "diff by ref" spec, so there are some differences.
> Those differences are the crux. If I go to the ISO standard, I don't  
> find an alternate list of elements. I find a document saying, in  
> essence, "this document creates a standard based on the W3C  
> specification, with some restrictions, which are listed here." That  
> isn't a competing specification, it's a standard based on a  
> specification.

AFAIK, the WHATWG is a proper superset of the W3C specification. So,  
it seems to be a specification based on a specification. It's not done  
by reference, of course. But I seriously doubt that's the rubicon.

> And you're quite right, the ISO document is the "standard," the W3C  
> document is the "specification." I used the word "standard" in my  
> earlier email in the casual sense that many use when referring to  
> well-established, normative W3C documents, but that's a bad idea in  
> this discussion.

I don't really care, myself.

>> If the WHATWG spec remains a superset, then I think the likely  
>> *technical* fragmentation reasonably can be seen as fairly minimal.
>> Whether there are significant social/marketing issues, well, I  
>> guess the real question is *how* significant they are. That I don't  
>> know.
> Nor I. But as an indication, when I first started delving into  
> HTML5, I was certainly confused by there being two widely-publicized  
> and linked "specifications" for HTML5. I've seen other people being  
> confused by that, and I don't think it's just that we're easily  
> confused people.

But people are often confused by there being multiple documents (i.e.,  
normative and non-normative). People are confused by what a working  
draft entails. Etc. The question is what's the significant harm of  
that confusion and how prevalent it is.

Since none of these are even recommendations, I think we should be  
rather relaxed about things, esp. as the content reasonably overlaps.

There's lots of confusion to be had as this is a multi-year (decade?)  
effort involving lots of things. Why is this confusion esp. significant?

> The W3C version needs to either be the spec, or not, and if it is it  
> needs not to be referring back to a different document also claiming  
> to be an HTML5 specification (whether or not it follows that with  
> "with additions").

But...are you confused now? It seems that the confusion is resolvable.

Note that the working group could, in theory, tomorrow, vote to go to  
last call with the WHATWG version spec. Nothing in the process forbids  
that it the least. It's obviously unlikely, given the players  
involved, but it's not impossible. Or even wholely unreasonable.

> For clarity, that's the one issue I'm weighing in on, as a member of  
> one of the large target audiences. I'm not commenting on (or  
> endorsing comments on) issues of copyright, licensing, or other  
> issues raised in this thread.


My only point is that multiple specs, even with the W3C stamp, exist  
all over the place and for a variety of reasons. They are sometimes  
confusing (I've reviewed working drafts only to find out that  
everything I found had been addressed in the editor's draft already).  
I also think that pushing this particular line may be, ultimately,  
counterproductive, and certainly not worth the gain. But, you know,  
it's all guessing on my part. What I can say is that I don't feel that  
the W3C needs to establish a principle of owning specs in this way, at  
least, all the time. There's a reasonable case to be made for that,  
esp. for *recommendations*.


Received on Thursday, 10 June 2010 09:17:01 UTC