W3C home > Mailing lists > Public > xml-dist-app@w3.org > October 2001

RE: Issue 4 Proposed Resolution (was: why no doc type declarati on and PIs in SOAP)

From: <Noah_Mendelsohn@lotus.com>
Date: Mon, 1 Oct 2001 17:26:49 -0400
To: "Champion, Mike" <Mike.Champion@SoftwareAG-USA.com>
Cc: xml-dist-app@w3.org
Message-ID: <OF754A6393.BC07D922-ON85256AD8.0073E7B5@lotus.com>
Mike Champion writes:

>> All I'm saying is that there is no equivalent 
>> way in to say "don't send me
>> no stinkin' PIs or DTD internal subsets," 
>> in a way that an XML parser can
>> enforce

You seem to be assuming that the only things that the specification for an 
XML application can rule out are things that a DTD or schema can check 
for.  Not so.  First of all,  I can write a math vocabulary including 
constructs such as


and my spec can say "legal documents will only have prime numbers in the 
<prime> tag.  That seemse to be a fine application-specific restriction on 
 XML, and is surely one that neither DTDs nor schemas can capture. 

Also,  nothing in XML requires you to use a DTD (or schema) even in the 
situations where it would express a restriction.  I can write a spec that 
says "don't use attributes", and I certainly don't have to write a DTD to 
say it.  There are many important XML vocabularies for which neither a 
schema nor a DTD is ever written, and that's OK if it meets your needs.

So, I don't see why every application of XML has to define a meaning for 
every construct that the XML specifications makes available, except to 
explicitly prohibit those that it does not allow.  I certainly don't see 
that the subset adopted normally will be one entirely expressible in a DTD 
or schema.  Each application should pick the XML features that it needs, 
being sure that the convention adopted is indeed a legal (and preferably 
tasteful) subset appropriate to its purposes. 

I agree with Andrew Layman that PI might be worth reconsidering, but we 
would have to be careful to define its significance.  Is it associated 
with a nearby header entry?  Can it affect SOAP processing?   In short, 
just allowing it adds some complexity.  In the case of DTDs, I really 
don't want to pay the implementation overhead of dealing with them.  SOAP 
is already being beaten on by those who say "I had a great binary protocol 
and your XML version sacrifices too much performance in the name of 
interop."  We can make SOAP perform if we try hard, but let's not make the 
problem unduly harder.  Thank you.

Noah Mendelsohn                                    Voice: 1-617-693-4036
Lotus Development Corp.                            Fax: 1-617-693-8676
One Rogers Street
Cambridge, MA 02142
Received on Monday, 1 October 2001 17:35:32 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 22:01:16 UTC