|
FIXML_Schema
< Previous Next >
Re: Validation via Schematron
Rob James / HSBC Bank plc <>
2 Oct 2006 4:57PM ET> > Hi Greg,
> >
> > I was a member of the ISDA/FpML Validation Working Group and also the
> > FIXML Schema WG. For FpML we established a set of business validation
> > rules and attempted to implement them in Schematron and CLIX. For our
> > purposes, Schematron was not able to handle the complexity of some of
> > the business rules for OTC Derivative Contracts, however CLIX was.
> > Areas where we ran into trouble for FpML usually involved verifying
> > intermediate payment date schedules given a start and end date.
> >
> > This may not be an issue for FIXML as listed assets are well specified
> > either by an exchange or the issuer. Determining that the proper
> > elements and attributes are included for various messages should be
> > straight forward. Actually verifying the data values for those
> > elements, or attributes, may not be. Schematron seems to fall short if
> > computation, or extrapolation, based on data is required.
> >
> > Robert Stowsky
> >
> > Co-Chair FIX GDC Technical Subcommittee
> >
> > > Has anyone had any experience using Schematron to validate FIXML
> > > messages? I am using Schematron in conjuction with XML Schema to
> > > validate algorithmic parameters of a message. But, I think it might
> > > be useful to expand this type of validation to the entire message.
> > > In other words, create a Schematron document that covers all the
> > > tags. Please chime in if you know of any traps/pitfalls regarding
> > > this method. Thanks. -Greg
>
> Greg,
>
> As the current chair of the FpML Validation Working Group I can re-
> iterate what Robert has said. As part of our mandate we are looking into
> a adopting new rule language to codify our semantic validation rules.
> Schematron is a possibility but as Greg points out this will only cover
> about 90% of rules. NRL is another possibility and I've had
> conversations with Christian Netwich about this. Another possibility is
> OCL which we will be evaluating at some point in the future.
>
> My personal belief is that none of these languages will ever be human
> readable and yet still expressive enough to define all rules one
> could dream up.
>
> I've talked to some of the ISO20022 guys about this as well and we have
> agreed to share the approach we take. I hope we can include FIXML in
> this collaboration as well.
>
> Cheers,
>
> Simon Heinrich Product Development Director, C24 Chair, FpML Validation
> Working Group
Simon,
We created NRL because OCL is not a business facing language. Whilst any machine processable language will not be poetic, NRL is very easily understood by business types.
Suggest you look at the examples Christian has created for FpML or the examples we have produced for the FIX Rep WG.
In either case the main issue with readability is the schema not the NRL grammar...
Should we consider combing the Algo and Rep WG work as we seem to be covering the same ground ?
Regards,
Rob
Re: Validation via Schematron
Rob James / HSBC Bank plc 2 Oct 2006 4:57PM ET
|
