[html4all.org] [HDP] Response to Review of HTML Design Principles Questionnaire (fwd)

[Leif Halvard Silli]

Rob,

On 2007-08-23 05:26:37 +0200 Robert Burns <rob at robburns.com> wrote:

> While I am concerned the way they apply these principles (and the way  they 
> equivocate everywhere even in the language of these principles),  I think 
> they honestly do not mean this the way you're interpreting  it. This 
> principle is not really related to the headers= issue.

I would say, unlike Lachlan, that it is related to the headers= issue. I do not say that to attack headers= is their point with this principle.

> It is  about the UA 
> conformance criteria, which insists that UAs continue to  support existing 
> content.
> 
> A better way to state this (with less of their masterful equivocating  
> language) would be "Strive to support all existing content".

This does indeed sound better. However, this is a 'spesification design principle'. As such, «strive to support» is good as «moral» for us, the spec-ers. But this principle rather describe a goal - about what should be spec-ed for UAs.

«Specify how HTML5-UAs shall support significant/all existing mark-up»

In my view, this principle is aimed at being able to declare HTML4 and below void and not relevant.

They could have said «All HTML5-clients must also support HTML4 and below + have defined behaviour for widespread proprietary extensons of those spesificatons».

As Lachlan stated it, it is not even certain that «support existing content» means «support it the way it is spec-ed in the specs», it seems rather to mean «support it as one have gotten used to».

> This way  it is 
> clearly stated as a goal and not sing the qualifier words like  "significant" 
> which they'll later reveal to mean not the sites you  think are significant.
> 
> So I think they would say that HTML5 will support headers= as a UA  
> conformance criteria. In other words all browsers will be expected to  

Well - I am not quite with you there. Well, it depends on what they say. One of the points with my reply was to tell that headers= can indeed, unlike what Lachlan said (about noise etc), be related to this principle.

> support the association between headers and data cells using the  headers 
> attribute (as well as other attributes and the new HTML5  algorithm). 
> However, they want to (or at least are considering)  dropping headers= 
> entirely from the document conformance criteria.  This means authors who 
> produce document s using headers= anywhere  will be non-conforming with 
> HTML5. They will get an error from the  conformance checker for every 
> headers= attribute in their document.

As this is a UA principle, we could say that HTML5 aims to create a [theoretical] «UA conformance checker» as well. And perhaps, by this standard, they could define headers= as «necessary to support». But a) they don't even support it today, so this would be an improvement! b) However, just as authors will not always care about conformance, neither will UA vendors - and if they don't see the point with headers=, they won't support it.

> So Lachclan's example of <marquee> where it is not in the document  
> conformance criteria (authors are not supposed to use it), however it  is in 
> the UA conformance criteria (meaning UAs are supposed to parse  it, add it to 
> the DOM and maybe provide some tacky styling to it; I  haven't read how much 
> support for marquee is in there)

I understand this.
-- 
leif