W3C home > Mailing lists > Public > public-html@w3.org > June 2010

Re: table layout conformance (was: some conformance questions)

From: Ian Hickson <ian@hixie.ch>
Date: Thu, 17 Jun 2010 22:38:18 +0000 (UTC)
To: Daniel Glazman <daniel.glazman@disruptive-innovations.com>
Cc: HTML WG <public-html@w3.org>
Message-ID: <Pine.LNX.4.64.1006172156520.19265@ps20323.dreamhostps.com>
On Thu, 17 Jun 2010, Daniel Glazman wrote:
> >
> > "Tables must not be used as layout aids. Historically, some Web 
> > authors have misused tables in HTML as a way to control their page 
> > layout. This usage is non-conforming" 
> Thanks for letting me know. I am then officially objecting to that last 
> sentence of this paragraph in the spec, and I want it to be removed.
> No conformance checker can test this so this cannot be a conformance 
> criterium. A content editing tool will never be able to enforce such a 
> criterium.

The specification makes a distinction between machine-checkable 
conformance criteria and non-machine-checkable conformance criteria. There 
are a number of the latter; the above is not the only one.

Conformance checkers are explicitly not required to check these 
conformance criteria, and editors are explicitly not required to enforce 
compliance to them. This is defined in the conformance section:


There are generally speaking two ways to approach the problem of people 
misusing markup: either define that their documents mean something 
different than they think it means, or define that using the wrong markup 
is incorrect. When advocating good markup use, in my experience it is much 
easier to convince an author if you can point to a place in the spec that 
outright says that their document is invalid than if you have to convince 
them that their document means something different than they think. The 
latter is too "meta" for many authors. This is why the specification has 
these requirements.

I think it would be reasonable to suggest that the spec should make it 
clearer which requirements are intended to be machine-checkable without 
user interaction and which today could only be checked by someone able to 
determine the author's intent. Labeling each such requirement would be a 
significant amount of work, but if you think it would be worth it anyway, 
I recommend filing a bug requesting this.

Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'
Received on Thursday, 17 June 2010 22:38:51 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 29 October 2015 10:16:03 UTC