Github user jcamachor commented on the issue:
https://github.com/apache/orc/pull/249
@omalley , it seems like a good idea, let me explore it and refresh the PR.
I will adapt HIVE-19226 to these new changes too.
@wgtmac , I understand you are suggesting that this can be fixed only from
Hive side? Problem is that existing ORC files should still be read properly,
hence you would need to recognize old vs new ORC files. In addition, you will
apply displacement twice when reading/writing, in Hive and in ORC. It seems to
me the cleaner solution is just being able to point to ORC that timestamp data
is in UTC from Java reader/writer. FWIW, change to stringify in
TimestampColumnVector is needed indeed.
---
