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

Multistructs - why not.

From: Jacek Kopecky <jacek@systinet.com>
Date: Tue, 18 Dec 2001 02:02:42 +0100 (CET)
To: <soapbuilders@yahoogroups.com>
cc: <xml-dist-app@w3.org>
Message-ID: <Pine.LNX.4.33.0112180149510.27474-100000@mail.idoox.com>
 Hi all. 8-)
 [CCing xml-dist-app but mainly targetting soapbuilders]
 Someone has mentioned multistructs here. Here's why I don't want
them in SOAP Encoding (this is reasoning, not a list of points):
 1) they are compound types with accessors distinguished by name
_and_ position, therefore structs and arrays are special cases of
 2) but arrays show that some more information is useful (item
type, array size) which is not present in multistructs, so it
would seem that arrays are not a special case of multistructs,
 3) there we have a contradiction so either of these cases is
 3a) multistructs can represent "simple" arrays where the
additional information above is not needed and therefore arrays
can be represented very naturally as multistructs containing
    i) at most one 'itemType' accessor,
   ii) at most one 'arraySize' accessor,
  iii) any number of 'member' accessors,
 3b) multistructs are meant to be able to represent every arrays,
therefore the additional information must be allowed there,
presumably on the first member accessor of its kind,
 3c) multistructs are in fact meant as structs containing arrays
and therefore should be represented as such;
 4) all 3x) choices mean more complexity in the spec unless we
just kick the term multistruct out of the spec, in which case 3c)
holds _and_ the spec is simpler.
 Please tell me if you cannot agree with this reasoning and why.

                   Jacek Kopecky

                   Senior Architect, Systinet (formerly Idoox)
Received on Monday, 17 December 2001 20:02:45 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 22:01:17 UTC