: Interesting question... I guess we haven't had one contrib depend on : another yet, or at least, I haven't checked to see if we have.
we do actually, a good example is xml-query-parser depending on the queries contrib. the current best practice for doing this is to set a property in your contrib build.xml based on the existence of the foo.jar you want, and have a "build-foo" target your other targets depend on that uses <ant antfile=.../> to build the foo contrib unless the property is set. then modify the appropriate classpath to include foo.jar. I would think the best way to handle benchmarks for contribs would be to do the above, but not make the build-benchmark target a dependency fo all of your targets -- just for a new "benchmark" target which executs the benchmarker using the appropriate classpath. if more then one contrib starts using the bencharking stuff, this can be refactored into contrib-build.xml. a slightly related suggestion: perhaps the benchmark contrib should be split up a bit so that the code for running benchmarks is more seperated from the specific code/conf/alg benchmarks of the lucene-core, which would also be seperated from the code/configs for fetching the various corpuses. then it would be easier to see easily how to use the benchmarking code without being distracted by the implementation of hte benchmarking system. (this really only matters if there is an interest in creating benchmarks for more then one contrib) -Hoss --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
