I disagree with this decision. For one you get a sense of what the trunk looks like now - an errored state needs to be fixed either by marking a test as ignored or reverting your changes until everything looks good again. A second point is you could hide other problems that happen during this accepted downtime, not everybody is working on this known issue, other failures can be introduced in the meantime.
To illustrate the second point see our current build failure, which is assumed to be about the permissiong WIP when in fact it is not: [ERROR] COMPILATION ERROR : [ERROR] /home/travis/build/apache/jackrabbit-oak/oak-it/mk/src/test/java/org/apache/jackrabbit/mk/test/BasongoMicroKernelFixture.java:[21,41] cannot find symbol symbol : class MongoConnection location: package org.apache.jackrabbit.mongomk.impl [ERROR] /home/travis/build/apache/jackrabbit-oak/oak-it/mk/src/test/java/org/apache/jackrabbit/mk/test/BasongoMicroKernelFixture.java:[40,12] cannot find symbol best, alex On Thu, Apr 11, 2013 at 10:12 AM, Jukka Zitting <[email protected]>wrote: > Hi, > > It looks like the Travis build is timing out in > org.apache.jackrabbit.oak.run.BasicServerTest. I'll look into that. > > Meanwhile I adjusted the build configuration to only notify oak-dev@ > when the build status changes from success to failure or vice versa, > instead of flooding us with failure notifications for each failing > commit. > > BR, > > Jukka Zitting >
