[GitHub] [geode] PurelyApplied commented on pull request #2593: GEODE-5850: Do not build jar in test.

2018-10-11 Thread GitHub
Upon immediate hindsight, this is absolutely not the right way to do this. Maybe examine the class, get a list of dependencies, and depend on those. As it stands, our `repeat-new-tests.sh` will run the entire `*Test` instead of just repeating the changed tests within that scope. [ Full

[GitHub] [geode] PurelyApplied commented on pull request #2593: GEODE-5850: Do not build jar in test.

2018-10-11 Thread GitHub
Similar to the other comment, there's no real reason this has to get packaged in the `jar` to the full `o.a.g.m.i.c.c` package. Would this test read cleaner if the jar put the `SerializableNameContainer` at the top-level package? I know I'm overthinking this. [ Full content available at:

[GitHub] [geode] PurelyApplied commented on pull request #2593: GEODE-5850: Do not build jar in test.

2018-10-11 Thread GitHub
Any opinions on where this data container class should live? I kept it on the path to be the same as the test that needs it, but that's not really necessary. It could belong to `geode-dunit` as a shared resource, if we track down other places that need a simple `java.io.Serializable` data