Re: Let's get some principles nailed down

Tim Bray suggests:

>> CP3. When using XML, designers SHOULD NOT introduce 
>> syntax constraints beyond those involved in the 
>> definition of XML.

This seems a bit vague.  Taken literally, it implies that every 
application must accept all vocabularies (purchasing system must accept 
sports scores), which is clearly not what you intend.  Even ignoring the 
absurd interpretations, it could be taken to imply that all vocabularies 
must make at least some use of attributes (since that's a syntactic 
element and not allowing them is a form of constraint.)  Also not what's 
intended.  Yet another interpretation might be:  all software (or 
hardware) that accepts or processes XML...it's not OK to write software 
that accepts only the XML for one application (and which, by the way, 
makes no use of attributes.)  Still not right.

I suspect there are some fairly limited rules that would rise to the level 
of at least SHOULD NOT.  For example, you might say:  when using XML, 
designers should not restrict the use of non-significant white space. 
Designers must accept as equivalent all constructions that XML says are 
equivalent, e.g. attr='xxx' and attr="xxx".  Having gone that far, might 
it make sense to say instead:

CP3Revised:  Designers of applications SHOULD when practical define 
semantics for XML applications in terms of the infoset, as opposed to 
depending on serial forms.  When using serial forms, any two documents 
that have the same infoset SHOULD be treated as semantically equivalent. 

(...and then we have to clean up the mess that we have both an Infoset and 
an XPath data model that are gee, almost the same but not quite.)

------------------------------------------------------------------
Noah Mendelsohn                              Voice: 1-617-693-4036
IBM Corporation                                Fax: 1-617-693-8676
One Rogers Street
Cambridge, MA 02142
------------------------------------------------------------------

Received on Thursday, 21 November 2002 22:58:49 UTC