Niels, it is interesting to see that in the Hadoop community that reading a
gzip file through from the beginning to the split point is worth it for
some real user scenarios.

[image: image.png]
*From: *Robert Bradshaw <[EMAIL PROTECTED]>
*Date: *Wed, May 15, 2019 at 5:52 AM
*To: * <[EMAIL PROTECTED]>
*Cc: *dev

Interesting thread. Thanks for digging that up.