Steve Loughran

Looks good. Be aware that in CDP we will not backport this without the logs disabled, unless/until we have our own code all happy, which will involve a move to...

oh, you are right. ready to go in...yetus is just complaining about deprecation. +1, merging

what is the status of this? can we get it into branch-3.3 for the next release?

just want to say this was a really amusing title in the git logs. might use it myself "replace NPE with working code in ..."

@jojochuang can you do a rebase and push this up so we can see what yetus says now?

would be good for some reviews here from @mukund-thakur , @mehakmeet and ideally @sunchao and @dongjoon-hyun -both of whom will be able to review the matching spark-side change, which is...

thanks. fixed tests, ran locally, and ran the abfs ITest subclass. all good

> esdk-obs-java:3.20.4.2 still depends on okhttp, at least that would restrict the use to a subset of the code, not hdfs

Still a WiP but should interest people working with abfs/gcs and to a lesser degree s3a

this isn't a bug, just inefficient, as the instanceof is always false if the argument is null. because this is generated code, it's not something which can be fixed. so...