----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/38079/#review97591 -----------------------------------------------------------
-1, don't review this - the generated code is broken. Will fix/update. - Bill Farner On Sept. 2, 2015, 6:08 p.m., Bill Farner wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/38079/ > ----------------------------------------------------------- > > (Updated Sept. 2, 2015, 6:08 p.m.) > > > Review request for Aurora, Kevin Sweeney and Maxim Khutornenko. > > > Bugs: AURORA-1476 > https://issues.apache.org/jira/browse/AURORA-1476 > > > Repository: aurora > > > Description > ------- > > We encountered this bug because tasks containing `null` collections would > return from the database with an empty set/list. We could debate over > whether `null` is preferable, but ultimately we need to force consistent use > of an empty collection or `null` to make this work across the board. > > The specific issue in this case is that we try to share relations to entries > in the `task_configs` table by querying to find a pre-existing entry with the > same contents. We use object equality to determine this. That approach is > problematic when the 'same' content from thrift is not `equals()`. > > TODO(wfarner): Add a test case outside the scope of storage code to point > specifically at the impedance mismatch that was addressed here. > > > Diffs > ----- > > src/main/python/apache/aurora/tools/java/thrift_wrapper_codegen.py > b5f2bc9e54b525a6a782d8873c9112f6496cd3f2 > > src/test/java/org/apache/aurora/scheduler/storage/AbstractTaskStoreTest.java > 295974a9f97e020dce11474d500a1bcd40d9f5d5 > > Diff: https://reviews.apache.org/r/38079/diff/ > > > Testing > ------- > > > Thanks, > > Bill Farner > >
