On Mon, Nov 29, 2010 at 3:45 PM, Sim IJskes - QCG <s...@qcg.nl> wrote: > River has a normal build, and a QA framework. This QA is a lengthy process > taking more than 22 hours, where the river networking environment is tested.
Sounds like the perfect argument for a dedicated slave :-) > Whe have to wait for a long time to see the failure results of our changes. > The categories bundle the single tests together in order to test 1 specific > aspect of river. We currently have 1496 test setups in 66 categories. Whe > have noticed that the different machines produce different failures. The > purpose of the matrix project type fits this setup perfectly. Well, of > course except for potential scalability problems. :-) Right, having 66 copies of the workspace on different machines sound less than optimal. Would it be possible to divide these categories in some other way in order to maintain one or a few workspaces? /niklas
