RE: [codelist-comment] Public Review of Codelist Requirements

["David RR Webber \(XML\)" <david@drrw.info>]

Tony, 

From the CAM (content assembly mechanism) perspective these use cases
fall into the area of business rules more than simple codelists. 

Obviously there are hybrid grey areas - where a rule is combined with
varying selections - in traditional EDI these are linked codelists - if
selection A,B or D then codelist values 1,2,3,4,5 else 1,5,6,7,8,9 -
etc. 

The examples cited below are often written as software rules in
implementation.  I'd expect the ability to associate rules with a
codelist - at minimum a business rule expressed in english - that is
logically complete and able to be re-written in machine instructions. 

So in the FpML example - this " e.g. the value must be in
the range 000 to 99999 OR must equal one of 199999, 299999, 399999 ...  
etc. Is there any suggested way to handle these types of enumerations, 
(other than entering every single value into the codelist) ? " 

I'd expect something like: 

Associated Business Rule: Code values must be in the range GTE 000 and
less than 99999. 
Associated Business Rule: Values must be of the format N99999 where N is
integer digit and 99999 is a code value. 

These can then be implemented as a combination of data picture masks and
explicit if ( ) clauses and / or substring() functions to test explicit
parts of the code.

As part of the codelist model - you could further regulate these types
of expressions and formalize how they are written - to ensure logical
deterministic results.  Maybe its enough to simply state that the rules
should result in deterministic complete outcomes - which is why people
use codelists in the first list - to avoid "unbound" results.

DW

 -------- Original Message --------
Subject: [codelist-comment] Public Review of Codelist Requirements
From: "Anthony B. Coates (Miley Watts)" <abcoates@mileywatts.com>
Date: Mon, November 06, 2006 3:17 pm
To: "Comments to OASIS Code List Representation TC"
<codelist-comment@lists.oasis-open.org>

I am posting this on behalf of Simon Heinrich of C24 and FpML.  Cheers, 

Tony.

-------- Original Message -------- Subject:  Public Review of Codelist
Requirements
Date:  Mon, 06 Nov 2006 19:15:56 +0000
From:  Simon Heinrich <simon.heinrich@c24.biz>
To:  codelist-comment@lists.oasis-open.org


In SWIFT FIN and in FIX I've seen several instances of enumerations  
dictating a
range of values OR one of an additional set of values, e.g. the value
must  
be in
the range 000 to 99999 OR must equal one of 199999, 299999, 399999 ...  
etc. Is
there any suggested way to handle these types of enumerations, (other
than
entering every single value into the codelist) ?

I see in the code list representation requirements v1.0, under item F .1
 
"It
should be possible to represent code lists that cannot be enumerated",
consideration has been given to some types of non enumeratable lists,
but  
that
does not cover this case explicitly.


Cheers,


Simon Heinrich
Product Development Director, C24
Chair, FpML Validation Working Group

-- 
Anthony B. Coates
Senior Partner
Miley Watts LLP
Experts In Data
+44 (79) 0543 9026
Data standards participant: ISO 20022 (ISO 15022 XML), ISO 19312,  
UN/CEFACT TMG, MDDL, FpML, UBL.

---------------------------------------------------------------------
To unsubscribe, e-mail:
codelist-comment-unsubscribe@lists.oasis-open.org
For additional commands, e-mail:
codelist-comment-help@lists.oasis-open.org