Community
Participate
Working Groups
We have a handful of tests that count certain objects in the repo in order to assert that the test passes. Some examples: PartialCommitTest.java ResourceTest.java PushTransactionTest.java BranchingTest.java InvalidationTest.java CommitInfoTest.java PackageRegistryTest.java What these tests do, is typically something like: 1. add some stuff to a resource 2. perform some operations 3. query the *entire* repo for some of "the stuff" 4. verify that the number of "stuff" is as expected The implicit assumption in (3) is that the repo/store was empty before the test, for otherwise the number of objects found cannot be predicted. This makes it impossible for the test to work with an existing store. (And it happens to be my aim to make this possible for as many tests as we can, in order to improve test performance.) Therefore, I'd like to modify these tests where possible to count more selectively, e.g. to query only into the test-specific resource instead of the entire repo. In some cases, the test logic might get too complicated if I change them in this way. For such cases, I'd like to introduce a method ConfigTest.needsCleanRepository(), that the test class can override to return true, so that the test infra knows not to re-use an existing store, if one is present.
> In some cases, the test logic might get too complicated if I change > them in this way. For such cases, I'd like to introduce a method > ConfigTest.needsCleanRepository(), that the test class can override to > return true, so that the test infra knows not to re-use an existing > store, if one is present. On second thought, an annotation on the method is obviously better, as the config is set up and torn down per method.
Removed 'blocks' relationship with bug 335653
Moving all open enhancement requests to 4.1
Moving all open issues to 4.2. Open bugs can be ported to 4.1 maintenance after they've been fixed in master.
Closing because of no activity on this bug.