Github user andrewor14 commented on the pull request:
https://github.com/apache/spark/pull/126#issuecomment-37484105
It's true that finalizers are not predictable, but what this patch is doing
is no worse than before. Cleanup() logic is decoupled from finalize(); it can
still be called explicitly elsewhere. It's just that finalize() adds a case in
which cleanup() is called (and this case is when the RDD / block goes out of
scope). Also, the "severe performance penalty" in the article refers to a
relatively small increase (on the
order of milliseconds), which is not really an issue since these clean-ups
don't happen very often.
---
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 infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
