[ubl-ndrsc] [Fwd: Re: [ubl-comment] To define propietary businessdocuments]

["Eve L. Maler" <eve.maler@sun.com>]

Uh, I think I kind of gave Gunther/Mike a head start here on the 
"identifiers as properties" paper.  As we discussed, we really don't 
need a paper so much as something we can submit as a comment on CCTS. 
What do people think of this attempt to describe the problem space? 
(Here I also describe what I think UBL ought to do about it, but I don't 
describe what I think *CCTS* ought to do.)

	Eve

-------- Original Message --------
Subject: Re: [ubl-comment] To define propietary business documents
Date: Mon, 01 Jul 2002 08:35:26 -0400
From: "Eve L. Maler" <eve.maler@Sun.COM>
To: Philip Goatly <philip.goatly@bolero.net>
CC: ubl-comment@lists.oasis-open.org
References: <5.1.0.14.0.20020630102808.03525060@popmail.cortland.com> 
<008201c220da$6561e8f0$f66ca8c0@bolero.net>

Hello all-- Now that the talk has turned to codes and identifiers, I'd
like to add one point.  Unfortunately, I will soon be going on a
week-long vacation and might not see any future additions to this thread
for a while...

Philip's description of the distinguished semantics of codes and IDs
below makes a lot of sense to me.  What I'd like to add is that
sometimes a piece of information has both semantics: a code can be used
as a unique identifier.  Likewise, a number, a text string, a URI
reference, or a name can be used as a unique identifier.  The ID aspect
contributes a higher-level usage semantic (e.g., "you'll want to use
this ID to retrieve the thing/meaning itself") and the
code/number/whatever aspect contributes the representation (e.g.,
"you'll want to format your ID this way").  In fact, without saying what
the representation of an ID is, there is no way to guess the right
answer, given all these choices.

What this boils down to, for me, is that the code/number/whatever should
be understood to be the *representation* of a piece of information that
serves as the identifier *property* of an object class.  This is not a
new idea, but I don't think anyone has formally made this suggestion on
the CCTS yet.  I would be very glad to see us apply this idea to our
modeling, because I think an answer pops out easily in each case.  In
practice, I think it means the Identifier Representation Term would
never get used, and that all identifiers would end up with UBL names
such as <something>IDCode and <something>IDNumber.

	Eve

Philip Goatly wrote:
 > Tod,
 >
 >    Codes are a shortened representation of something
 >
 >    e.g US - for the united states, M for male
 >
 >     ID - is a piece of data that uniquely identifies a particular 
instance
 > of something or someone .
 >
 >      e.g U167166111B - to identify a particular container
 >            on a ship
 >           or MM-123-666 to identify a box by its mark or number.
 >
 >       An ID can be fairly abitrary, and transitory where as a code 
within a
 > a particular list is unique,  unchangeable universal.
 >
 > my 2ps worth
 >
 > Cheers, Phil

-- 
Eve Maler                                        +1 781 442 3190
Sun Microsystems                            cell +1 781 883 5917
XML Web Services / Industry Initiatives      eve.maler @ sun.com