Re: On HTML5 conformance and conformance checking

On Mon, 14 Apr 2008, Steven Faulkner wrote:
> 
> So while the web may one day contain many documents that get the HTML5 
> conformance checker "stamp of approval" It will actually be littered 
> with non conforming HTML5 documents..

This is already the case even with machine-checkable conformance criteria 
today. The HTML4 validator on validator.w3.org issues "stamps of approval" 
even for pages that are non-conforming for a wide variety of errors, for 
example href="" attributes with syntactically invalid URIs.

And even if that was fixed, it wouldn't check that the document uses 
elements in a conforming way (e.g. that <table> elements aren't used for 
layout, which is invalid in HTML4, or that <h1> elements are used for the 
correct outline level, or that table cells don't overlap).

Futhermore, even if we required the alt="" attribute, we would _still_ not 
check the conformance of the alt attribute -- there is no way to tell, 
from a conformance checker, whether the alt="" value is appropriate or 
not.

So this is unrelated to whether alt="" is required in all cases or not.

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

Received on Monday, 14 April 2008 19:46:25 UTC