W3C home > Mailing lists > Public > www-xml-schema-comments@w3.org > April to June 2000

Re[2]: XML Schema Structures

From: Ivan Kurmanov <ivan@tm.minsk.by>
Date: Mon, 12 Jun 2000 17:43:30 +0300
Message-ID: <16738.000612@tm.minsk.by>
To: Noah_Mendelsohn@lotus.com
CC: www-xml-schema-comments@w3.org, Thomas Krichel <T.Krichel@surrey.ac.uk>
Dear Sirs,


Noah Mendelsohn wrote:

Nlc>From the description of Model Group [1]:

Nlc>         (all, (in which case {particles} is restricted 
Nlc>         to contain local and global element declarations, 
Nlc>         with {min occurs}=0 or 1, {max occurs}=1) 
Nlc>         contain exactly zero or one of each element 
Nlc>         specified in {particles}. The elements can 
Nlc>         occur in any order. 

Nlc> So, I believe that we have exactly the facility that you are requesting. 

Thank you a lot.

It solves a part of a problem. But still it is not exactly the facility
that we need. A problem which remains has not been explicitly stated in my
first message: it is to allow that some elements (inside 'all' model group)
can be repeated twice or more. In other words, we would like to have
limitation {max occurs}=1 to be removed. We understand that it is a bit
late now because the document's last call period has expired.

When I say "we" I mean the electronic library project of RePEc [1]. We
rely heavily on humans to maintain and supply the library with
bibliographical data. Until now we used a our own data format, heavily
influenced by the IAFA template. We need a simple data format for
humans to create and maintain data. And now we a going to migrate to
XML and to use XML Schemas to define a data language. Please, consider
the following data example:

<paper >
       <title>...</title>
       <author>
               <name>Adam Smith</name>
               <email>...@...</email>
               <homepage>http://...</homepage>
               ....
       </author>
       <abstract>...</abstract>
       <file> .... </file>

       .... <!-- many other elements can appear here -->

</paper>


This is how a simple piece of data may look like. But for the 'paper'
element the following restrictions apply:

 - 'title' element is required to appear exactly once
   {minoccurs} = {maxoccurs} = 1

 - 'abstract' element is optional, but cannot be repeated
   {minoccurs} = 0, {maxoccurs} = 1

 - 'author' element must appear at least once, but may be repeated
   {minoccurs} = 1, {maxoccurs} = unbound    !!!!!

 - 'file' element is optional, but may be repeated
   {minoccurs} = 0, {maxoccurs} = unbound    !!!!!

All elements can appear in any order.

Our point shall be quite clear by now: we do not see a reason to limit
{maxoccurs} to be equal to 1 in {all} model group.

That is what we need and below I'll try to explain why. 

We want such data to be created and kept up-to-date by many of our
partners. But we cannot require or assume that all they necessarily will
use schema-driven XML editors, we want them to be able to create
valid data files even in a plain text editor. The reason for this is that
our electronic library is for academic research (primarily in economics)
and it is a free library available on the net. We and our data
maintainers providing free on-line publications. And our data maintainers
come from more than a hundred of different institutions. Everybody works
in a unique environment.

So we require as little as possible from them. To remember that a
'title' element is always required for a paper, but only one 'title' can
exist, and that a 'name' element is required for a person, but only one 'name'
is OK. These requirements are obvious. But we can't require them to
remember  some  artificial  required  order which can be described by the
sequence {compositor} in a schema.  And we still need some elements to be
repeatable.

If you look around, most objects when described with their relationships
to other objects do have unordered sets of properties. In a bit
simplified, generalized view of the world the properties of objects of a
particular class are usual to always have a single value, some usually
may have several values, and some may have no value at all.

I personally think that this kind of set of properties is very common in
the real world. Therefore if XML Schemas will allow to define such sets
of properties, it will allow to define easy to use data languages with
obvious structural restrictions. They will be obvious because they will
be based on the real-world relationships.

We ask the empowered parties to consider to remove the {maxoccurs} = 1
restriction from the all ModelGroup description of XML Schema: Structures
draft.


Thank you.


[1] RePEc: Research Papers in Economics
    http://netec.mcc.ac.uk/RePEc/


Ivan Kurmanov <ivan@tm.minsk.by>
Received on Monday, 12 June 2000 10:45:57 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 23:08:47 UTC