Community
Participate
Working Groups
Having bug 365444 fixed we could consider running a random set of Git commands (with JGit API) against a test repo and compare the result with CGit output. Once the test fails, the set should be remembered to serve as repeatable steps, a test case. The trickiest part here seems to be comparing with CGit which is not always available. Moreover, this kind of test can fail in the least expected moment, and it doesn't have to be related to the latest fix. Thanks to Dawid for his inspirational talk on RandomizedTesting[1] at EDC in Poznan[2]. [1] https://github.com/carrotsearch/randomizedtesting [2] http://wiki.eclipse.org/Eclipse_DemoCamps_November_2011/Poznan