RE: FW: XHTML 1.1 and DOCTYPE declaration

There are many descriptions on the web of how DOCTYPE switching affects layout (based on hits for "DOCTYPE switching"). Existing UA's that might support strict (vs quirky) rendering based on DOCTYPE switching might not be implemented to also switch based upon the document's default namespace.

I'm not against removing superfluous text from the minimal valid XHTML document, and, if the DOCTYPE requirement limited the potential of XHTML 1.1 in some way, then I'd understand the desire to remove it. However, I do think it's valuable to maintain compatibility with existing UAs that render XHTML as 'strict' HTML.

--Brad

-----Original Message-----
From: Steven Pemberton [mailto:steven.pemberton@cwi.nl]
Sent: Thursday, March 29, 2007 7:37 AM
To: Shane McCarron
Cc: Brad Pettit; w3c-html-wg@w3.org; XHTML WG
Subject: Re: FW: XHTML 1.1 and DOCTYPE declaration

On Thu, 29 Mar 2007 15:15:32 +0200, Shane McCarron <shane@aptest.com>
wrote:

> The namespace does not identify the markup language.

It's true that it doesn't identify the markup language, but it does
distinguish XHTML from tag soup, which is what Brad was asking for.

The Schema group's position is I think that XHTML Basic is a subset of
XHTML 1.1 *and* a subset of XHTML 1.0 strict *and* a subset of XHTML 1.0
transitional, and so a document that conforms to XHTML Basic can be
reasonably validated with any of those schemas; it therefore need not be
the document's job to say what it is (I have the same problem myself with
documents saying which stylesheet they use; I often wish I could say
http://example.com/style.css?http://example.com/home.html for example)

Steven

> All of our various markup languages use the same namespace, for good or
> for ill.  SVG had it right; multiple namespaces for multiple languages.
> But we dont have that option.
>
> Steven Pemberton wrote:
>>
>> On Wed, 28 Mar 2007 22:31:06 +0200, Brad Pettit
>> <Brad.Pettit@microsoft.com> wrote:
>>> Requiring the DOCTYPE makes it more straightforward to differentiate
>>> XHTML when there is no http content-type available. Otherwise it could
>>> appear like quirky HTML.
>>
>> I partly agree, but wouldn't the namespace be enough? No one is going
>> to put the namespace on quirks HTML surely?
>>
>> Best wishes,
>>
>> Steven
>>
>>>
>>> -----Original Message-----
>>> From: w3c-html-wg-request@w3.org [mailto:w3c-html-wg-request@w3.org]
>>> On Behalf Of Shane McCarron
>>> Sent: Wednesday, March 28, 2007 12:07 PM
>>> To: HTML WG
>>> Subject: XHTML 1.1 and DOCTYPE declaration
>>>
>>>
>>> At the risk of reopening something better left closed...
>>>
>>> We have from time to time discussed the wisdom of requiring or not
>>> requiring a DOCTYPE declaration on XHTML Family Documents. Note that
>>> M12N itself does not say anything about this, deferring instead to the
>>> conformance requirements of markup languages defined using M12N.
>>>
>>> In the most recent public XHTML 1.1 draft
>>> (http://www.w3.org/TR/xhtml11/) we indicate that the DOCTYPE
>>> declaration
>>> is required.  In the current editors draft
>>> (http://htmlwg.mn.aptest.com/htmlwg/xhtml11) we have changed this to a
>>> should (this was done on 20 February by meprobably when dealing with
>>> the
>>> XML Schema group's comments from ages ago).
>>>
>>> I have been thinking about this A LOT this past week, and I feel we are
>>> making a serious mistake here.  This is our markup language, and agents
>>> need a way to know what language they are encountering when they see
>>> documents in OUR language.  the DOCTYPE declaration is the only
>>> portable
>>> mechanism we have for declaring this right now.
>>>
>>> I understand the arguments about XML Schema, and I don't care.  If you
>>> want to validating using a schema implementation, go ahead.  Use the
>>> schemaLocation attribute to point to our schema implementation.  But if
>>> you happen to point to a local copy or something, how is a user agent
>>> to
>>> have a CLUE about what markup language you pretend to be using?  I
>>> suppose we could require that if a schemaLocation is used it MUST point
>>> to our well known location... at least that way a user agent author
>>> could do a mapping.
>>>
>>> Anyway... I think we have made a mistake here.  There is no reason I
>>> can
>>> think of to make the DOCTYPE declaration a should instead of a MUST.
>>> Having a DOCTYPE declaration does not mandate any special form of
>>> processing nor of validation as far as I know.   Finally, removing the
>>> requirement for a DOCTYPE declaration is a major change for XHTML 1.1.
>>> If there are agents that expect this requirement from the original
>>> version of  XHTML 1.1 they will not correctly process documents from
>>> this new version.
>>>
>>> Comments?
>>>
>>> -- Shane P. McCarron                          Phone: +1 763 786-8160
>>> x120
>>> Managing Director                            Fax: +1 763 786-8180
>>> ApTest Minnesota                            Inet: shane@aptest.com
>>>
>>>
>>>
>>>
>>
>>
>

Received on Friday, 30 March 2007 02:48:52 UTC