W3C home > Mailing lists > Public > public-html@w3.org > February 2009

Re: What's the problem? "Reuse of 1998 XHTML namespace is potentially misleading/wrong"

From: Maciej Stachowiak <mjs@apple.com>
Date: Mon, 16 Feb 2009 11:23:31 -0800
Cc: Lachlan Hunt <lachlan.hunt@lachy.id.au>, HTML WG <public-html@w3.org>
Message-id: <52966ED6-EC90-41A1-8995-CD8C8C64971D@apple.com>
To: Larry Masinter <masinter@adobe.com>

On Feb 16, 2009, at 11:10 AM, Larry Masinter wrote:

> I'm glad we agree on the principles, but I don't think we agree on  
> the remedy.
> This relates to ISSUE-4 ("html-versioning") and ACTION-93:
> It seems like we have four versioning mechanisms for content:
> * MIME type
> * namespaces
> * version attribute on elements
> All of these have been rejected as ways of distinguishing XHTML5  
> from XHTML2,
> for various reasons which have been analyzed:
> http://www.w3.org/QA/2007/05/html_and_version_mechanisms.html
> Only the namespace element is sufficient for distinguishing languages
> in arbitrary generic XML  compound documents where there may not be  
> even
> the possibility to specify the HTML version.
> So, I offer the following proposal:
> HTML5 isn't being designed for incorporation in arbitrary generic  
> XML compound
> documents in any case.

That would be an incorrect assumption. XHTML5 will be likely be used  
in compound documents in combination with, for example, SVG or XBL2.

> So I would propose instead:
> a)  XHTML5 *MUST NOT* be used in arbitrary generic XML compound  
> documents. Any
>   use of the namespace in such contexts denotes XHTML2, not XHTML5.
>   Since HTML5 and XHTML5 are not being designed for inclusion in  
> arbitrary
>   XML compound documents, this should not have any impact in practice.

Since this proposal is based on a false premise I do not think it is  
viable. Further, since user agents in practice support XHTML namespace  
elements in arbitrary XML compound documents, this would amount to  
requiring XHTML5 user agents to implement XHTML2, which is not a  
reasonable requirement.

> b) a new MIME type, application/xhtml5+xml, be allocated for XHTML5,  
> and a new
>   MIME type, application/xhtml2+xml, be allocated for XHTML2, since  
> they both
>   differ sufficiently from XHTML1.

HTML 4.01 differs from HTML 3.2 and XHTML 1.0 yet all three can be  
served with the text/html MIME type. More importantly, though, user  
agents will likely process application/xhtml+xml (and indeed  
application/xml or text/xml) documents under XHTML5 processing rules  
rather than XHTML2 or XHTML1, so the new MIME type won't provide  
meaningful value.

>    Since recommended practice is to deliver HTML5 as text/html in  
> any case,
>   this should not have any serious impact in practice either.

I would not assume what has impact in practice based on what is  
recommended. In practice, many things are commonly done that are not  

I prefer your original proposal to close this issue without action.


> Larry
> --
> http://larry.masinter.net
> -----Original Message-----
> From: Maciej Stachowiak [mailto:mjs@apple.com]
> Sent: Monday, February 16, 2009 10:43 AM
> To: Larry Masinter
> Cc: Lachlan Hunt; HTML WG
> Subject: Re: What's the problem? "Reuse of 1998 XHTML namespace is  
> potentially misleading/wrong"
> On Feb 16, 2009, at 10:21 AM, Larry Masinter wrote:
>> I think I wasn't clear enough, so let me try to clarify:
>> In XML, namespaces identify vocabularies, not languages.
>> Language definitions define rules for putting together
>> vocabularies and user-supplied terms and procedures by
>> which a receiver of an utterance in the language (a
>> receiver) can interpret and understand the meaning
>> intended by the sender of that utterance.
>> You can have two different languages sharing the same
>> namespace as long as there is sufficient information
>> for any receiver of an utterance to determine which
>> language was intended.
>> Conclusion:
>> XpMLq and XrMLs can share the same XML namespace
>> as long as there is some way that you can determine
>> which was intended, through MIME types, DOCTYPE
>> or other means.
>> This is true whether XpMLq and XrMLs are XHTML1
>> and XHTML2, XHTML2 and XHML5 or XHTML5 and
>> XHTML5.001 (an incremental update).
>> Sharing namespaces with different languages is
>> unacceptable if there is no other versioning
>> mechanism than the namespace name. It would
>> define two overlapping languages where there
>> is no deterministic way of deciding which
>> utterances were in which language, without
>> extra contextual information. This is very
>> bad language design.
> XHTML2 and XHTML5 both use the application/xhtml+xml MIME type, and do
> not require a doctype declaration to be specified. XHTML2 has a
> version attribute on the root element, but it is optional.
> Furthermore, either may be used in a generic XML compound document
> where there may not be even the possibility to specify the HTML
> version via a root element attribute or doctype declaration. In
> practice user agents assign semantics and behavior almost exclusively
> based on the qualified name of an element, since there is no other
> disambiguation or versioning construct that is present in all  
> contexts.
> Thus, I believe the use of the same namespace by XHTML2 and XHTML5
> would be "unacceptable" and "very bad language design" by the standard
> you set out above.
>> Whether or not there is any "vendor" who supports
>> or is intending to support both XHTML2 and XHTML5
>> is irrelevant to this question, since agents
>> that cannot properly interpret an instance
>> of the language they're presented with should
>> not blindly try to interpret those instances
>> in some other language.
> I was trying to say politely that XHTML2 is unlikely to ever be widely
> implemented by mainstream Web software, so we shouldn't spend a lot of
> the group's time worrying about ways it may conflict with XHTML5. If
> someone were to invent a language that uses all the same words as
> English but with different meanings, then in theory that would be bad
> language design but that doesn't mean English speakers need to worry
> about it a whole lot.
> To summarize, I definitely think there is a problem in theory, for the
> same reasons you gave above. But I don't think it will be a problem in
> practice.
> Regards,
> Maciej
>> Larry
>> --
>> http://larry.masinter.net
>> -----Original Message-----
>> From: Maciej Stachowiak [mailto:mjs@apple.com]
>> Sent: Thursday, February 12, 2009 4:19 PM
>> To: Larry Masinter
>> Cc: Lachlan Hunt; HTML WG
>> Subject: Re: What's the problem? "Reuse of 1998 XHTML namespace is
>> potentially misleading/wrong"
>> On Feb 11, 2009, at 12:46 PM, Larry Masinter wrote:
>>>> But XHTML2 also has several major incompatibilities with XHTML1,
>>>> which
>>>> would effectively make it impossible to implement both XHTML 1.x
>>>> and 2
>>>> in the same implementation, if they share the same namespace [3].
>>> The fact that one language makes <title>abc</title> equivalent
>>> to <meta property="title" content="Document Title"/> and the other
>>> does not does mean the languages are incompatible. But the elements
>>> "title" and "meta" have the same meaning as vocabulary items, their
>>> usage is different in the two languages.
>>> I know that there is significant resistance to the idea that you
>>> might
>>> define a vocabulary independent of a language which uses that
>>> vocabulary,
>>> or that you might define a language independent of the processing
>>> rules
>>> by which instances of text in the language should be processed, but
>>> those separations are fundamental to the design of XML, and the
>>> question
>>> here is about XML namespaces and their use.
>> What Lachlan raises is an actual technical problem in the design of
>> the languages, even if the languages have "the same vocabulary" in
>> some sense. It is not possible to implement a user agent that  
>> conforms
>> to both XHTML5 and XHTML2. In many cases the same elements or
>> attributes are defined to have different processing requirements.
>> Since DTDs are optional, and since elements in the XHTML namespace  
>> may
>> appear in a compound document, there is no way to determine which
>> processing requirements to use. XHTML2 also has this problem with
>> regard to XHTML1.x.
>> I don't believe any vendor currently plans to support XHTML5 and
>> XHTML2 together in the same product, so perhaps this is not a problem
>> in practice. But it is still a design flaw in XHTML2 and should be
>> fixed by the XHTML2 Working Group.
>> Regards,
>> Maciej
Received on Monday, 16 February 2009 19:24:20 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 29 October 2015 10:15:42 UTC