Github user viirya commented on the pull request: https://github.com/apache/spark/pull/3600#issuecomment-66435647 I agree with you that the saved operation here is a cheap one. :-) However the problem you mentioned would not happen with current version of `DeserializationStream`. Not all InputStream close their underlying stream when they are collected by GC. There are detailed discussions [here](http://www.coderanch.com/t/278165/java-io/java/InputStream-close-garbage-collection) and [there](http://stackoverflow.com/questions/1522370/does-input-outputstreams-close-on-destruction). I am sure that `FileInputStream` implements `finalize` to close underlying file. But other streams used here are not as the tests show. `DeserializationStream` is implemented in Spark and it has no such behavior. During modifying the codes, I checked it and found that you must explicitly call its `close` to close its underlying stream. That is why it passes the tests. I am ok to close this PR if it causes problem. But if it would not really cause the mentioned problem, I can not see why a slightly improved performance is bad.
--- 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. --- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org