- From: Jacek Kopecky <jacek@systinet.com>
- Date: Wed, 2 Jan 2002 19:44:27 +0100 (CET)
- To: Noah Mendelsohn <noah_mendelsohn@us.ibm.com>
- cc: "peter.hendry" <peter.hendry@capeclear.com>, xml-dist-app <xml-dist-app@w3.org>
Noah, at first it was not clear to me that you're reacting to my post, so maybe in some future post you might indicate that in a way. Thanks in advance. 8-) Back to business: I may be underestimating the value of datatypes in XML Schema, as I see that language mainly poised for validation. I'm nervous about using that language for description of data structures, in my opinion it's too powerful for that task, with its <all>s and <sequence>s, with its <choice>s and whatnot. AFAIK XML Schema is not even able to describe all the possible structures in the XML data model (a tree), only a somewhat restricted subset. I don't think going further would be practical for XML Schema. Anyway, I'm not against using XML Schema simple data types, on the other hand I think that higher-level data structures are too application-specific. We can agree that NameValue pair (or KeyValue) is used often where SOAP Encoding is used but I don't see in which way this struct is special and so why it should be a part of SOAP Encoding. Best regards, Jacek Kopecky Senior Architect, Systinet (formerly Idoox) http://www.systinet.com/ On Fri, 21 Dec 2001, Noah Mendelsohn wrote: > > Whatever the implications for the SOAP design, I think you are somewhat > underestimating the reason for builtins such as positive integer in the > schema language. Schema is NOT just used for validation. It is equally > important as input to tools that do mappings to database languages, > programming languages, etc. For these purposes, having well known and > agreed upon names for commonly used types is extremely important. > Otherwise, recognizing such types becomes a theorem proving exercise rather > than a simple recognition of type names (e.g. you would have to prove that > the facet restrictions on integer actually resulted in positive only. If, > for example, someone restricted the lexical space to forbid minus signs, > that might do it, but I bet it would be a mess to detect.) > > I think there is an analogy for SOAP encoding. Where we standardize well > known type names, tools are more likely to be able to generate effective > programming language mappings automatically. On the other hand, having too > many such well-known types makes the spec too big, and tends to get us in > the business of defining types that are more special purpose. I think the > NameValue and NameValueList types under discussion are in the grey area > where you can make a good case either way. Thanks very much. > > ------------------------------------------------------------------------ > Noah Mendelsohn Voice: 1-617-693-4036 > Lotus Development Corp. Fax: 1-617-693-8676 > One Rogers Street > Cambridge, MA 02142 > ------------------------------------------------------------------------ > > > >
Received on Wednesday, 2 January 2002 13:44:35 UTC