Re: [ubl-ndrsc] Containers

[Chin Chee-Kai <cheekai@softml.net>]

On Tue, 27 May 2003, Dan Vint wrote:

>>It seems like the wrapper element should still be used. If it is good to
>>put one in the container, then it is good to put them all in the container.
>>I have the feeling this "container" is going to be automatically generated
>>rather than specified, which is where the problem is coming from with EMPTY
>>elements.

Does your container contain multiples of same <item>s or multiples
of different <item>s?  I thought the discussion has been on containing
multiple occurences of same <item>s.

But I think you touched on an important point about containers and 
elements being EMPTYiable or otherwise.  For 1..n contained items
not all of which can have EMPTY content and nil attributes, 
a container of 1..1 is a straightforward answer, if containers
are used.  But when contained items themselves can all 
simultaneously have vanished data, then a 1..1 container 
doesn't invalidate anything but only make the instance space 
look "weird", as if there's an empty trashcan with a lot of
compartments.  If that doesn't  happen too often, I suppose it's 
not going to break things and  thus will not present too much of
an issue.



>>If the items to be contained 0..N, then the container should  become 0..N
>>(or better yet 0..1) and the items would then be 1..N. I believe the
>>results are the same and you still get to use the container in all cases -
>>which I believe should be required.

If an item has multiplicity 1..N, then it is containable and the
container is 1..1.  Think that was the NDR conclusion earlier on.

The challenge comes from the other case:  Given that the item has
multiplicity 0..N, should the container be 0..1 or 1..1?



Best Regards,
Chin Chee-Kai
SoftML
Tel: +65-6820-2979
Fax: +65-6743-7875
Email: cheekai@SoftML.Net
http://SoftML.Net/