Github user mridulm commented on the issue:
https://github.com/apache/spark/pull/19294
@szhem Did you resolve issues with newTaskTempFile, newTaskTempFileAbsPath,
etc potentially still throwing NPE due to path being null ?
I saw multiple invocations in spark sql which are calling into it ?
(`FileFormatWriter.newOutputWriter` for example)
I did run against SHC (spark hadoop connector) and that seems to work fine
for basic writes - though it might probably be because it was not exhaustive
and does not exercise other options in spark sql.
Essentially, any use of `path` in `HadoopMapReduceCommitProtocol` can be
potentially problematic if it can be `null`.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
