-Mark
Tim O'Brien wrote:
+1 on the idea of having these classes available for use, +1 on *temporarily* adding this to commons-math, but -1 to adding them to MathUtils (this code should be stored in a separate contrib source tree). Basic Financial utilities would add desired functionality, but in the "long term" - specific areas of application should not be a part of commons-math. I'm not saying this will never be the case, read on...
Alternatively, application areas should be separate modules entirely
(right now, within a separate contrib directory) - we don't want to
create a monolithic collection of code (or a monolithic community of
developers), but in the short-term I can see that adding these limited
utilities would encourage active community participation. In other
words, +1 for this if and only if we promise to keep these things
limited to very simple applications/algorithms.
I'd rather not see us start to get into too much specialization (not that I'm not interested) - IMO commons-math is similar to commons-digester. Commons Digester provides a basic mechanism for executing a set of rules, these rules are triggered by the structure of an XML file. Because commons-digester remains focused on a very limited set of goals, it is a utility which can be used for a wide range of applications - from marshalling/unmarshalling to/from XML, to a basic MathML engine, One could even write some sort of programming language using Digester..... The point is that Digester is a very generic piece of code. Likewise with the Linux kernel. Strength through simplicity.
We should define the limits of the project and resist the urge to exceed those boundaries. This will lead to more nimble and focused communities, and it will reduce risk. I've been following the discussions surrounding the Struts 1.1 RC2 release, and I think that it might be wise to draw bold boundaries around a project. Commons-math is X, and if anything falls outside of X we can add it as a contrib, but I don't want to open up a Pandora's Box on a sandbox component that hasn't made it through the promotion process yet. Also, if a package in contrib isn't passing 100% of the tests, it isn't a blocker for a release, AND we could distribute those classes separately (think ant.jar and optional.jar in Apache Ant).
If someone starts offering Apache utilities for Financial calculations, then why not start adding code for: Radioactive Decay, Monte Carlo simulations, Utilities for the calculation of gambling odds, Circuit simulation, Aerodynamics, Optimization algorithms for the layout of a Printed Circuit Board, Quantum Physics, Calculations performed in Maritime Navigation, etc....
The goal is to release a simple, focused commons-math - get something
through promotion, and either at promotion time or afterwards a
conversation should be started about future direction. I think we've
got a good amount of community-building to get through first. Add the
code to a "contrib" source tree.
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
