FpML Issues Tracker
closed
Crash
Always
Architecture
Admin
andrew
Summary
There is no clear distinction in the XML Shchema between the definition of something and the identification of something.
A good example of this is Party at /FpML/party. In the schema it says "A type defining party information.". In the FpML manual it says: "...this component is restricted to party identification.". These two statements are contradictory. Either Party defines the Party and all its data, or it is used to identify a Party. It cannot do both.
We need to clearly demarcate in the schema the difference between "definition" and "identification" and "reference". This is an architecture problem. Definition is all the data contined within the type. Idenitification is all the data used to identify the type. Reference is usually an intra-document link to a thing.
We need an architecture statement. My suggestion is that we establish to base types of "Identifier" and "Definition" and all types extend them appropriately.
Notes:
andrew
05/30/07 9:44 am
I can not see that this proposal provides any additional clarity to the model. The business terms used in FpML are clear enough to the majority of its users.
matthew
05/30/07 11:11 am
This needs to be discussed at the AWG rather than people posting their personal views.
What is “clear enough” for a small software company, may not be clear to an organization with 10,000s of people around the world implementing it.
matthewdr
03/06/08 2:59 pm
Suggestion at the AWG from Andrew Jacobs to use the following convention:
definition = no suffix
identification = “Id” suffix
reference = “Reference” suffix
matthewdr
03/06/08 3:00 pm
The AWG agreed to document this within the “Naming” section of the Architecture document.
mgratacos
03/06/08 3:03 pm
Just to clarify, when we talk about references, we mean intra-document references (id/idrefs).
mgratacos
03/20/08 1:10 pm
Action on AJ to add it to the Naming section.
andrew
05/21/08 8:45 pm
Section 2.2.4 was added to describe the proposed naming convention.
mgratacos
06/16/08 3:40 pm
The Architecture Spec 2.1 Recommendation includes the proposed naming convention as Andrew indicates.