Closed geoHeil closed 5 years ago
Regular spark 2.3.2 works fine, regular spark 2.2.1 also fails with above exception.
When creating a custom build like https://github.com/geoHeil/flint/tree/flint-spark-2.2 the basic example works again with spark 2.2
However, there are 2 unit test failures
- should correctly convert SQL TimestampType with default format *** FAILED *** (237 milliseconds)
[info] 1199232000000000000 did not equal 1199228400000000000 (CSVSpec.scala:100)
NOTE: csv test failure is also is on regular master branch
- partition preserving something failed because a list was empty and `head` was called
Hi!
The current flint version only supports Spark 2.3. On Sun, Oct 14, 2018 at 7:54 AM geoHeil notifications@github.com wrote:
When creating a custom build like https://github.com/geoHeil/flint/tree/flint-spark-2.2 the basic example works again with spark 2.2
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/twosigma/flint/issues/53#issuecomment-429620214, or mute the thread https://github.com/notifications/unsubscribe-auth/AAwbrBao_wLlK5O8IK28fqQZ26J2RywOks5ukyXjgaJpZM4XasUA .
val tsRdd = TimeSeriesRDD.fromDF(dataFrame = df)(isSorted = true, timeUnit = MILLISECONDS)
throws
on spark 2.2 when trying to create the initial RDD.
Minimal reproducible sample:
on spark 2.2 via HDP 2.6.4