Github user squito commented on the issue:
https://github.com/apache/spark/pull/23058
> causes any performance degradation compared to memory mapping
@ankuriitg good question, though if you look at what the old code was
doing, it wasn't memory mapping the file, it was reading it into memory from a
regular input stream, take a look at
[`ChunkedByteBuffer.fromFile`](https://github.com/apache/spark/blob/fa0d4bf69929c5acd676d602e758a969713d19d8/core/src/main/scala/org/apache/spark/util/io/ChunkedByteBuffer.scala#L192-L212)
basically doing the the same thing this is doing now, but without the extra
memory overhead.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
