Re: Intent to Conform (was Re: Version information)

On Apr 12, 2007, at 9:37 AM, Chris Wilson wrote:

> Laurens Holst [mailto:lholst@students.cs.uu.nl] wrote:
>> Chris Wilson schreef:
>>> All,
>>>         I need to detail an essay about compatibility and opt-in to
>>> explain the Microsoft viewpoint on this.  I have a few things on my
>>> plate I must do today, so it will probably take a day or two.  I'm
>>> going to likely be silent until then on this topic.
>>
>> I wonder how much merit it has to discuss/decide this now; isn’t the
>> necessity of this depending on the output of this working group and
>> the opinion of implementors on the feasibility to implement it  
>> without
>> breaking current pages?  Assuming on beforehand the spec will  
>> never be
>> able to define HTML5 in a way that doesn’t break existing pages is I
>> think premature.
>
> The current HTML5 spec has severe breakages from IE's  
> implementation already.  (E.g. removing the classid and codebase  
> attributes from <object>.)  This is why I needed to write up my  
> thoughts.

I'm not sure how this is a breakage. IE could process those  
attributes and still be conformant. But content could not use them  
and remain conformant. Nothing in the spec says you can't have  
extensions.

(I don't know the reasoning for this decision but I imagine it is  
because classid and codebase are generally used to request plugins  
that use a Microsoft-specific technology and only work in IE.)

>
>> You mentioned HTML6; if HTML6 is not interoperable without breaking
>> backwards compatibility, a version switch can be added in HTML6.
>> Saying that HTML5 won’t have versioning information does I think
>> not prevent future specifications from not having any.
>
> And saying that the DOCTYPE is <!DOCTYPE html5> rather than <! 
> DOCTYPE html> doesn't say that the next version actually needs to  
> be HTML6, or that it will necessitate a new DOCTYPE version.

Fair enough, but <!DOCTYPE html> has the significant advantage that  
it triggers standards mode in current browsers, so it's easier to  
make content that degrades gracefully. I don't think <!DOCTYPE html5>  
would.

And conversely, saying that HTML5 documents start with <!DOCTYPE  
html><html> (which is already distinguishable from HTML4) would not  
rule out requiring <!DOCTYPE html><html version=6> for HTML6.

Regards,
Maciej

Received on Thursday, 12 April 2007 17:55:56 UTC