If we parametrize the test categories to run (already done that in my local copy of the trunk) we could invoke ant like this:
ant -Dqatests.categories=id,jeri test.qa and be able to specify different categories for the two builds Most of the categories take pretty long to run, but some do not, like "id" for instance. I have run them all by now so I could compile a list of categories to run for each build. 2009/4/1 Jukka Zitting <[email protected]> > Hi, > > On Tue, Mar 31, 2009 at 5:03 AM, Jonathan Costers > <[email protected]> wrote: > > I notice Hudson is still using the make files to build the River QA > harness > > and tests (when I look at the console output). Maybe we should > reconfigure > > it to use Ant for building and running the harness and tests? > > Yes, that would be good. > > Do you have suggestions for the appropriate ant targets to invoke in > the CI build? Ideally we could have one reasonably fast CI build that > we can run as often as possible (i.e. whenever a commit is made), and > another more intensive CI build that's run daily or weekly depending > on how long it takes. > > BR, > > Jukka Zitting >
