OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help



   Re: [xml-dev] Interesting pair of comments (was Re: [xml-dev] Schema Ex

[ Lists Home | Date Index | Thread Index ]

----- Original Message From: "Elliotte Harold" > Pete Cordell wrote:
> You're mixing syntax and semantics. Handling <<<<< properly is a syntax 
> issue. It is a well-formedness error. Handling <p>This is 
> <strong>very</strong> important</p> when the schema says mixed content is 
> not allowed is a very different issue that requires a different response.
> regardless, mixed content is not as uncommon or unexpected as many people 
> think. It is not an accident. It is not bad form. It is not something to 
> be avoided. It is the very natural way to express many extremely common 
> constructs when modeling information, including so-called data-oriented 
> applications (as if any information content were not data). Excluding 
> mixed content from a schema language is like excluding a reverse from a 
> car's transmission. 99% of the time you might drive  in forward mode, but 
> not having the reverse gear when you need it will leave your car (or 
> application) stuck in a dead end it can't be extricated from without a tow 
> truck.

Well, I'm afraid I still disagree.  Whether it's a well-formedness error, or 
a higher layer semantic error, it's still an error in that domain.  If an 
application domain never needs a particular construct (and that condition 
might be circular in argument [1], but that's a separate issue), why does a 
language specifying syntaxes for that domain require the ability to specify 
that construct?  X86 processors support BCD arithmetic - that used to be 
core functionality! - , but neither Java, C#, C, or C++ support it natively.

That said, I would like to make it clear that I'm not arguing against 
supporting mixed.  More that if I had to choose between mixed and xsi:nil, 
I'd probably go with xsi:nil.  (although having both would be the best). 
Which is really just to emphasise what's core and what's not is quite 
difficult.  Maybe a true core would be useless to anyone, and only after 
suitable add ons would a profile become useful.  Maybe its all too 


[1] e.g. this specification language does not support X, because X is never 
used, because the specification language does not support X!
Pete Cordell
Tech-Know-Ware Ltd
                         for XML to C++ data binding visit
                         (or http://www.xml2cpp.com)


News | XML in Industry | Calendar | XML Registry
Marketplace | Resources | MyXML.org | Sponsors | Privacy Statement

Copyright 2001 XML.org. This site is hosted by OASIS