Hi all,
Mark proposed moving "taseree" as an experimental "drools-solver"
module
into the drools trunk. There a couple of things that might need a decision:
1) Does no one object to moving into drools as an experimental module?
It would not be included into the <module> set of the parent pom (yet),
as it's still very experimental.
But it will use the parent pom, so it might require additional
dependencies in dependencyManagement.
Currently it depends on commons-lang, commons-logging, log4j and
xstream. And drools of course :)
2) What name do we give it?
- taseree (original name, legacy)
-1 as its no longer correct as it's more then a tabu search with rule
engine evaluation: simulated annealing, etc
- drools-solver
+1 sounds good, although people might think it's an alternative to
for example ilog solver, while it's in a quite different league:
localsearch instead of simplex. Someday drools might build a simplex
based solver too...
- drools-solver-localsearch
+0 some of the code is already a bit reusable between other types of
solvers, such as genetic algoritms, so calling it only localsearch
doesn't really fit
3) What is it's package?
- org.drools.solver ?
3) How should the modules be moduralized?
trunk/ (parent pom)
trunk/drools-core
trunk/...
trunk/drools-solver
trunk/drools-solver-examples (depends on solver)
trunk/drools-solver-benchmarks (depends on examples)
Or
trunk/ (parent pom)
trunk/drools-core
trunk/...
trunk/drools-solver (solver parent pom)
trunk/drools-solver/drools-solver-core
trunk/drools-solver/drools-solver-examples (depends on solver-core)
trunk/drools-solver/drools-solver-benchmarks (depends on examples)
4) Code format
Currently I am using the standard "Sun's code conventions for Java",
much like JBoss hibernate, Jboss Seam, etc.
http://java.sun.com/docs/codeconv/html/CodeConvTOC.doc.html
However as a drools module it should use drools's code format of course.
What do you think?
--
With kind regards,
Geoffrey De Smet