Re: let authors choose text/html or application/xhtml+xml (detailed review of section 1. Introduction)

Hi Phil,

On Aug 31, 2007, at 12:22 PM, Philip Taylor (Webmaster) wrote:

>
> Robert Burns wrote:
>
>> I still don't understand how we disagree here. For any  
>> specification the issue of what is or is not valid has to be a  
>> part of the conversation. Disabling null end tags in a  
>> hypothetical HTML 4.02 DTD makes the use of <br /> valid HTML 4.02  
>> (and valid SGML). It's also already valid XHTML 1. So I'm not sure  
>> where we have a disagreement here. Are you just saying you don't  
>> want null end tags disabled in HTML? Then that's just another part  
>> of the language that we discuss, deliberate and work out (the same  
>> as deciding whether it should have a P element).
>
> I don't think it's really a matter of what I want, it's
> more an issue of the logic (or lack thereof) behind it.

It certainly matters what you and I and any of the members of the WG  
want. NET's can be disabled in SGML. When a DTD does so, its still  
SGML its just an SGML application that cannot use NETs. So the issue  
in building a specification (and especially if that specification  
included a schema defined through an SGML DTD), is do we want NETs  
enabled or disabled. Its the same as the question of do we want to  
have a global class attribute or not. Its all part of defining what  
is valid and what is invalid HTML (or in the case of NETs what is a  
valid or invalid HTML serialization).

> Ever since HTML was created (as a dialect of SGML),
> NETs have been a part of the language.

Yes, they have been a part of the language that an application of  
that language (such as HTML 4.01) could elect to use or disable.

> The fact that
> few if any browsers actually implemented them is neither
> here nor there : documents that used them correctly
> were valid (modulo other things), those that used
> them incorrectly were invalid (-- ditto --).

But if there was an HTML 4.02 that disabled them, or an HTML-4.01- 
strict-no-nets DTD that disabled them then the documents that used  
NETs would be one kind of valid HTML 4.01 and the documents that  
didn't would be another kind of valid HTML 4.01.

> Now there is a suggestion that NETs could
> be removed w.e.f. *HTML 4.02 (which isn't within our
> aegis, strictly speaking), but to what end ?  To allow
> documents that were previously invalid to become valid,
> if the DOCTYPE werechanged ?  What sort of reason is
> that ?  Or is it more that some XHTML documents could
> then be valid HTML documents ?  Again, to what end ?

I don't know to what end. It could provide a way for authors to  
support legacy HTML, XML and legacy SGML processors (if that's  
something anyone's interested in). Some of those other ends you  
listed seem fine to me. It would also allow authors to author with  
XHTML 1 syntax change to an HTML 4 DTD and still pass the validator  
and the scrutiny of those who might misconstrue "/>:" as a NET.

> Even worse (assuming that browsers do not alter their
> behaviour based on the DOCTYPE, much as I would like
> them so to do), you would end up with a situation in
> which an HTML 4.01 document using <meta ... /> in
> the head region would be flagged as "invalid" by
> the validator, whilst an *HTML 4.02 document with
> exactly the same markup would be flagged as valid,
> yet both would render identically in an HTML 5 compliant
> browser.  Could we really explain that to the
> web-page-author-on-the-number-57-Clapham-Common-omnibus ?!

We could just explain that for processing UAs that do make use of  
SGML DTDs the different doctype declarations would make a difference.  
If you want to enable the continued support for SGML applications,  
this would be one way to do that.

Take care,
Rob

Received on Friday, 31 August 2007 18:10:01 UTC