proposal for issue 164: describing arrays in XML Schema

 Hello all, 8-)
 according to my issue #164 [1] it is felt that we should say how
the SOAP Encoding and SOAP Data Model constructs are to be
described using XML Schema.
 (Just to let you know, the short version is that I'm proposing
to keep the status quo. The full proposed text is at the end.)

 With regards to this issue we have three options:

 a) we can create our own schema language (a very simple one,
actually) that would be used to describe the data structures,
everybody using a different language (like XML Schema) would have
to define the mapping to our schema language.
 b) we can define the mapping ourselves, keeping the 'SOAP Data
Model schema language' only implied.
 c) keep status quo and leave the mapping unspecified.

 The pros and cons of the various options:

 PRO a: a simple schema language in IDL-like languages (e.g.
WSDL) resulting in much simpler implementation and understanding
of the schemata;
 CON a: some work needed, not so much (I volunteer, BTW);
 CON a: one more language to deal with;
 PRO b: reusing XML Schema language, fewer languages to deal
with,
 CON b: uncleanliness of the solution (inappropriate reuse);
 CON b: some work needed, more than above (for covering the
complexities of the language may not be fun),
 PRO c: the least work;
 CON c: allowing incompatible mappings and confusion in
understanding of some of the nuances of XML Schema(*).

 Now, I have two real cases where going with a or b would help or
would have helped:
 Array description in WSDL might have not been so ugly (see [2]);
 XML Schema choice usage in WSDL debate would have been avoided
(see [3] and the following thread, esp. [4]).

 The ugliness of array description was mainly caused by the
compound attribute arrayType which we have rid ourselves of. The
current array serialization should be describable by XML Schema
using only default or fixed attributes in subtypes of enc:Array,
and no special attributes like wsdl:arrayType should be
necessary.
 The misunderstanding of XML Schema choice usage in WSDL is more
of an issue with WSDL because it's too vague about the meaning of
'encoded' use of its and XML Schema's constructs. So when the
next version of such a language is about to happen, hopefully
I'll be there to shout loud and jump high.
 So to reinstate, since we've chosen this path many times (and
it's the XML path), we can rely on the world sorting this one out
for us and itself. I propose that the issue #164 is closed with
the text, if it's still true by the time we get to it:

"The WG has decided to keep the status quo as the issue is
lessened by the resolution of issue 144 and the expected course
of WSDL resulting thereof; and no other substantial need for
in-spec specification of the mapping of SOAP Data Model schema to
XML Schema has been presented."

                   Jacek Kopecky

                   Senior Architect, Systinet (formerly Idoox)
                   http://www.systinet.com/

[1] http://www.w3.org/2000/xp/Group/xmlp-issues.html#x164
[2] http://lists.w3.org/Archives/Public/xml-dist-app/2001Oct/0171.html
[3] http://list.systinet.com/l/showmsg.xp?ln=systinet&mid=1985
[4] http://list.systinet.com/l/showmsg.xp?ln=systinet&mid=2178

Received on Monday, 17 December 2001 20:50:35 UTC