Andrea Aime wrote: > Ben Caradoc-Davies ha scritto: >> Andrea Aime wrote: >>> Did you spot what exactly is not working, which spi resource is not >>> being properly read? >> org.geotools.data.DataAccessFactory >> org.geotools.filter.FunctionExpression >> >> which both have main versions and test versions. The test ones cannot go >> in main because they reference test classes, and (if I recall correctly) >> cause factory loading to fail. > > Right. There is actually an open jira about the inability to have > separate output directories for src and test: > http://jira.codehaus.org/browse/MECLIPSE-422 > > I can see two paths: > - move back the eclipse output trick in a profile > - have the tests that need to register factories avoid using META-INF > and roll a custom factory iterator instead. As far as I can see > only three modules make use of test META-INF stuff: 3) revert back to the old way :)
Seems like 1) move the new eclipse config to a profile seems the nicest way. It gives developers the choice of how to set up their eclipse environment. > > find . -name services -type d | grep src/test > ./modules/unsupported/app-schema/app-schema/src/test/resources/META-INF/services > ./modules/unsupported/wfs/src/test/resources/META-INF/services > ./modules/library/main/src/test/resources/META-INF/services > > The second of course is bound to fail in other ways due to > other kinds of overwrites, but seems well worth the effort if > the module maintainer is using eclipse due to the reduced build times. > > Opinions? > > Cheers > Andrea > > -- Justin Deoliveira OpenGeo - http://opengeo.org Enterprise support for open source geospatial. ------------------------------------------------------------------------------ _______________________________________________ Geotools-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/geotools-devel
