#general
- e
enough-analyst-54434
08/18/2015, 12:57 AMbut breaking it was a big ongoing contest internally while they were hardening whatever they were using for a technique - r
rough-minister-58256
08/18/2015, 1:00 AMnice - h
happy-kitchen-89482
08/18/2015, 5:17 AM@witty-crayon-22786: Yes, having JVMDependencyAnalyzer be its own task makes perfect sense - h
happy-kitchen-89482
08/18/2015, 5:17 AMalthough shouldn't we not need it any more if we get rid of the global strategy? - j
jolly-chef-92794
08/18/2015, 2:51 PMI’m working on a patch (rb 2651) to make junit tests to run with same jvm as the
version they are compiled to, as per their-target
. To me it makes sense to have this enabled as the default option, but I currently have it turned off by default in case it would break/inconvenience others. Preferences?platform - a
ancient-stone-50795
08/18/2015, 2:54 PMWhat we want to make sure happens in our CI environment is if code is specified to compile with
that it runs with a java 7 JRE (if it is available). That way we can detect problems with the wrong bootclasspath, etc. We’ll still have the option even if the default is ‘false’, but it will be a knob you have to turn in order to get what I consider to be the most desirable behavior.-target 7 - a
ancient-stone-50795
08/18/2015, 2:55 PMIf you didn’t have a Java7 JRE then you could flip this switch off and run all of your java 7 tests with a java 8 JRE for example - g
gorgeous-fish-11040
08/18/2015, 3:37 PMwow. broke master. fixing - g
gorgeous-fish-11040
08/18/2015, 3:38 PMseems flaky - g
gorgeous-fish-11040
08/18/2015, 3:38 PMconnection error - g
gorgeous-fish-11040
08/18/2015, 3:40 PMGoFetchIntegrationTest.test_go_fetch_integration - g
gorgeous-fish-11040
08/18/2015, 3:45 PMRestarted and now CI is green 💚 - b
bored-art-40741
08/18/2015, 3:45 PMIs that fetching from an external source? - g
gorgeous-fish-11040
08/18/2015, 3:45 PM@adventurous-twilight-44110 @witty-crayon-22786 FYI ^^ - g
gorgeous-fish-11040
08/18/2015, 3:45 PM@bored-art-40741: it does - b
bored-art-40741
08/18/2015, 3:46 PMYeah, oh well. Cost of integration tests that aren't fully self-contained, but then, that's the point of integration tests - g
gorgeous-fish-11040
08/18/2015, 3:47 PMAgreed. But it was scary. Don’t like to break CI 😄 - g
gorgeous-fish-11040
08/18/2015, 3:48 PMmaybe add a retry to go_fetch.py? @witty-crayon-22786 @adventurous-twilight-44110 - g
gorgeous-fish-11040
08/18/2015, 4:05 PMretries for resolve.go https://rbcommons.com/s/twitter/r/2652/ - w
witty-crayon-22786
08/18/2015, 4:33 PM@happy-kitchen-89482: we have @adventurous-twilight-44110 working on dependency scoring, which has a lot in common with the dep analyzer - w
witty-crayon-22786
08/18/2015, 4:34 PMhis plan is probably a shared subclass with helpers for consuming that product - w
witty-crayon-22786
08/18/2015, 4:34 PMwith the existing analyzer as one task (in compile, to still fail the build on error), and the scorer in its own goal - w
witty-crayon-22786
08/18/2015, 4:36 PMthe only thing that people might still want from the analyzer after global is gone would be the check for direct deps - w
witty-crayon-22786
08/18/2015, 4:36 PMtransitive is enforced - w
witty-crayon-22786
08/18/2015, 4:36 PMand the scorer will get you unused (a 0% edge is unused) - h
happy-kitchen-89482
08/18/2015, 4:39 PMinteresting - h
happy-kitchen-89482
08/18/2015, 4:39 PMwhat's the purpose of dependency scoring? - w
witty-crayon-22786
08/18/2015, 4:40 PMif all 564 dependees of a target A use 1% of target A, then target A is badly factored - w
witty-crayon-22786
08/18/2015, 4:41 PMif only a few dependees B, C, D (of 564) use small amounts of target A, then it's possible that B, C, D are badly factored - h
happy-kitchen-89482
08/18/2015, 4:42 PMInteresting