Re: [COAS-List] Reminder: Conference Call

["David W. Forslund" <dwf@lanl.gov>]

At 02:02 PM 11/3/98 -0800, Tim Brinson wrote:
>"David W. Forslund" wrote:
>> 
>> As a follow up on the conference call today, I wanted to indicate an area
>> where XML could be of benefit.
>> Being able to add traits to PIDS dynamically could be very valuable.
>> Currently it requires adding some IDL
>> and recompiling and some programming.  If these traits could be specified
>> with XML they could be added dynamically
>> quite easily without having to reprogram the server.  We have generated a
>> DTD which describes an extended
>> set of traits as an example of what could be done.  This results in a
>> "template" of traits which could be used to
>> easily extend PIDS without reprogramming the IDL.  I think such a feature
>> could be of value for COAS, too, and
>> perhaps even more important.
>
>Sorry Dave I don't follow.  The traits in PIDS are name/value pairs
>where the name is a "string" and the value is an "any".  Why do you need
>to change any IDL to add new traits?  If the data type passed in the any
>is a new data type then you would have to create the IDL for that.  The
>data types for the HL7 and vCard traits are all strings right now.  When
>you get the string it has to be parsed by the appropriate rules.  If you
>receive XML it would also be parsed - by XML rules instead of HL7 ASN.1
>or vCard.  For HL7 V3.0 I would expect PIDS traits will be XML.  Are you
>suggesting it should use XML for all traits?

Sorry for my lack of clarity. 
I'm suggesting XML for a data type passed in the Any.  We could have an IDL
for these XML datatypes.
 and as you say it appears that HL7 will suport this in V3.0.  But if I
want something more than HL7 I should
still be able to define it in XML and thus the parsing would be handled by
XML (e.g. your Cinderella Trait 
should be dynamically defineable).  Now if I want to have some other
structure in my Any, I have to define the 
IDL for it.  I would like something more like LDAP, which allows one to
extend the datatypes handled without 
having to build new code.  In LDAP one can define some objectClasses which
do this. It would be most appropriate 
to do define extensions in XML rather than the way that LDAP does it.  Thus
the extensibility would be through XML 
(although one could still do this the way it is done now).  This would
allow one to publish to a Trader, say what traits 
you support and where one would look to parse them.  I think this
capability will be even more important in COAS
than it is in PIDS.  Perhaps this is what you mean by templates?
>
>The reason name/value pairs were chosen are for the dynamic
>capabilities.  Traits are tagged data.  XML is also tagged data.  Are
>you saying we should have tagged tagged data?

I'm not sure.  But I would like to be able to dynamically create more
complex traits such as your Cinderella Traits 
without having to do any programming or rebuild a server.  

Is this any clearer?
>
>
>Tim
>
>