The knob based approach falls more in line with what Spark users would expect. 
If they write native spark with an HDFS input like 
`sc.textFile("hdfs:///input/war-and-peace.txt")`, they wouldn't expect the file 
to be broken up differently depending on the number of cores they run with.

[ Full content available at: https://github.com/apache/beam/pull/6181 ]
This message was relayed via gitbox.apache.org for [email protected]

Reply via email to