Github user willb commented on the pull request:
https://github.com/apache/spark/pull/189#issuecomment-39812228
I'm still tracking down exactly where this problem is coming from, but
here's a little more detail on what's going wrong with Python accumulators when
the closure passed to `runJob` is captured. In this case, `CompletionEvent`
objects have `accumUpdates` fields that map from accumulator IDs to empty
lists. As far as I can tell, the same byte arrays for accumulator updates are
getting passed over the wire in both cases.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
