FpML Issues Tracker
closed
Minor
Have not tried
Validation Rules
Admin
danieldui
Summary
I propose we introduce the principle that each rule must have a single context. If a rule currently has multiple contexts then it must be split into multiple rules.
If a rule has multiple contexts, then it is merely coincidence that it works across both structures. If it works across multiple contexts because of a common supertype, then the rule should be moved to the supertype instead.
Relying on this coincidence doesn't work well for statically typed rule languages - such as OCL and NRL. XPath and XQuery cope well, but rule languages tend not to support this usage.
I am sure Daniel and Christian would have particularly useful comments on this.
Notes:
matthewdr
07/21/09 1:53 pm
Discussed at the VWG.
Daniel proposed that we should check the examples (e.g. eqd-26), to see if there should be a common supertype or be split into two or more rules.
Agreed to establish a principle that each rule should have a single context. This should be thoroughly enforced for version 5.
Agreed to formulate a spreadsheet of all the examples and decided for each whether there should be a common supertype or two rules.
matthewdr
07/28/09 1:22 pm
Discussed at the VWG today. Irina agreed to produce the spreadsheet showing multiple context and classifying whether to split each or create a common supertype.
iyermakova
08/07/09 8:52 pm
Added a spreadsheet showing multiple contexts and classifying whether a rule has multiple contexts or merely coincidence. VAL WG will review the list and determine whether it can be split or create a common supertype.
danieldui
03/29/10 3:21 pm
Discussed on meetings on 2010-03-09 and 2010-03-23.
The group agreed to keep rules as they are: With multiple contexts. FpML 5.0 should solve the problem. Decided to make support for multiple context a requirement for revised rule language.
danieldui
03/29/10 3:31 pm