The fussy part of me wants to say that this is a different bug: the scanner tests don't leak any more.
All of the connections in the puller acceptance tests are done in subprocesses, I think. Does the test framework catch lingering processes?
The fussy part of me wants to say that this is a different bug: the scanner tests don't leak any more.
All of the connections in the puller acceptance tests are done in subprocesses, I think. Does the test framework catch lingering processes?