Re: QASG last call comments: Modesty requirement

On Thu, 3 Feb 2005, Gary Feldman wrote:
> > 
> > Because whether they comply is not cut and dry. For example, SVG 1.2 
> > claims to comply to AWWW. But IMHO it doesn't.
> 
> Since the AWWW document explicitly disclaims having any conformance 
> requirements, it's not a specification.  So that's not a compelling 
> example.

Ok, HTML4. It says:

   HTML 4 is an SGML application conforming to International Standard ISO 
   8879 -- Standard Generalized Markup Language [ISO8879].

...yet the truth of that statement is debated.


> In my opinion, if compliance is not cut and dry, then that's a failure 
> of the specifications.

The technologies that we're talking about here are _very_ complex. There's 
no chance that any of them will ever be perfect -- just look at how issues 
are raised about SVG, CSS, or HTML. CSS, for example: CSS2 came out in 
1998, seven years ago. The CSS working group spent some three years 
fielding issues with CSS2, and addressed every single one, creating 
CSS2.1. There were over 300 issues recorded during the Last Call stage of 
that process alone. Since CSS2.1 went to Candidate Recommendation there 
have been more than 130 further issues raised, many of which actually go 
all the way back to CSS1 (1996, nine years ago). Claiming conformance to 
_CSS_ is therefore a bit rash.

Now, Unicode, QASG, Charmod, and other specifications that other 
specifications are likely to claim conformance to are maybe not on that 
level of complexity, but I still don't see that it is sensible to claim 
conformance to them. Why would anyone _want_ co claim conformance, anyway? 
Shouldn't conformance be something that the reader should determine of his 
own volition? I can understand claiming that an attempt was made to 
conform, but whether the editors were successful or not is not something 
the editors should determine, IMHO.

(Much like a spec can say it was designed with the intention of being 
simple and backwards compatible, but whether it actually is or not is not 
something the editors should make claims about, at least not in the spec.)

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

Received on Friday, 4 February 2005 14:04:27 UTC