RE: [ciq] Use of OASIS code list specs. in CIQ TC V3.0?

-------- Original Message --------
Subject: Re: [ciq] Use of OASIS code list specs. in CIQ TC V3.0?
From: "Ram Kumar" <kumar.sydney@gmail.com>
Date: Tue, April 17, 2007 1:44 pm
To: "David RR Webber (XML)" <david@drrw.info>
Cc: "Max Voskob" <max.voskob@paradise.net.nz>, "Bruehlmeier,David"
<david.bruehlmeier@sap.com>, "David Bruehlmeier"
<typo3@bruehlmeier.com>, ciq@lists.oasis-open.org,
colin_wallis@hotmail.com

Hi David,

I am glad that you now understand where I am coming from. We have NZ Government, who is an immediate user of this approach. I have asked

Ken to help to modify the xNL and show how the codelist approach can

be used. I hope he will be able to assist that will enable me to speed up

the process of including the codelist approach in CIQ, unless someone

in our TC is willing to do this work.

Regards,

Ram

On 4/18/07, David RR Webber (XML) <david@drrw.info> wrote:

Ram,

OK - I'm with Ken on this one!

Having it all in the one schema is a bad idea. You need to be able to versioning and do subsetting based on context.

And so the two stage approach is the right one IMHO.

Notice - I'm also following his genericode approach with CAM. I run an xslt against his genericode list (.gc) - to produce the CAM sparse lookup codelist.xml - to get best performance / context at runtime.

So - what Ken is saying is point to the schema for the codelist - and not import it in CIQ - because its not designed for that anyway. If you did import it into schema - it would have to be as a tokenized list of values.

Notice that approach fails for cross-codelist lookups - e.g. if countrycode = "UK" then currencycodes="[list1]" - or more likely - if countrytypecode ="EU" then currencycodes - "[list2]" - where the values of one list - down select against the other list.

More reasons to do a 2 stage pass - otherwise you get caught - where the schema either allows content thru as valid that is not - or gives an error when its OK.

Also - what about multiple encodings? Often the list changes - e.g. if 'US' - allow imperial - else use metric codes.

And this is what the W3C is enabling anyway - simple tokenized lists is OK for boolean or short static choices - otherwise - you should supplement the XSD with tools like CAM and schematron.

Errrh - where does this leave you? UBL already have the genericode approach - so they will use that with CIQ.

Are there other customers we need to provide for right now - or do we wait until someone has a use case for us?

DW

"The way to be is to do" - Confucius (551-472 B.C.)

ciq message