|
Greetings, PRWG folks. It’s been a long time since we’ve been active in
this group, because no one has raised any issues that required our
attention. (There have been a few PRWG related issues raised against the
FpML standard, but these have been able to be dealt with without the whole
group). Now there is an issue that requires us to take a look at
something. There has been an issue raised about our use of references
in the PRWG area. In some cases (e.g. the “objectReference”
element) we have references to objects that can be of a variety of types. For
example, we can have a valuation that references either a trade, a trade leg, or
a basket constituent. The problem is that for some tools (e.g. binding
frameworks), it can be difficult to process this part of the schema if the type
of the target is unknown. One solution might be to create an abstract
base type for all of the potential targets, but this is not necessarily
practical. (For example, trades, tradeIds, products, legs, and basket
constituents would all have to have the same base type). The other solution is to change the existing single element reference
(objectReference) to a choice of several elements, with each element pointing
to a different target type (or base type). For instance, you could have a
choice of tradeReference, legReference, constituentReference, etc. The second approach has been recommended by the Modeling
Task Force. While I’m not too keen on it personally, it does seem
like it may be the best solution to the problem. If adopted, this change would happen in 5.0. Does anyone have any comments on this proposed change? Regards, Brian Lynn |