Re: [ubl-dev] Customisation of specialised types

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

Perhaps one easier way which does not need modification to 
schemas nor fudge with NDR is to just use existing IdentifierType.
This sets your system to accept more IDs than what you would
consider valid, but would filter off IDs which you would 
also consider invalid.  So at this point, the invalid ones relative
to UBL are taken care of without rejecting what you might have 
considered valid, but there are those invalid ones relative to 
your system which are accepted as valid by UBL's IdentifierType.
(If you read about the hypothesis tests in statistics, this is
like their Type I & Type II errors).

At this point, just do an application-specific validation routine
to take as input the data accepted under UBL's IdentifierType,
pick it apart to see if it conforms to the string-integer mix
format in the sector you have in mind, and then decide again 
whether to accept or reject the datum, and/or as a result, the
entire document.



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


On Tue, 5 Jul 2005, Fraser Crichton wrote:

>>Hi,
>>
>>I have a dumb question.
>>
>>I'm interested in creating a customised Identifier type (a combination
>>of a string and a number), the sector refers to this type as a number
>>throughout all their business processes and systems and want to see it
>>referred to as a Number in any XML messages.
>>
>>Would I create a new specialised datatype and extend from the
>>udt:TextType? Would I create a completely new type based on xsd:string?
>>How else could I achieve this ugliness and comply with the NDR?
>>
>>Cheers,
>>
>>Fraser
>>
>>--
>>Fraser Crichton
>>XML Developer
>>SolNet Solutions Limited
>>L12, SolNet House, 70 The Terrace
>>PO Box 397, Wellington, Aotearoa / New Zealand
>>http://www.solnetsolutions.co.nz
>>DDI: 04-462-5078
>>Mob: 027-278-3392
>>Fax: 04-462-5011
>>