[
https://issues.apache.org/jira/browse/HBASE-30100?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Junegunn Choi resolved HBASE-30100.
-----------------------------------
Resolution: Fixed
> Fix misleading Javadoc for Table.batch about null in results array
> ------------------------------------------------------------------
>
> Key: HBASE-30100
> URL: https://issues.apache.org/jira/browse/HBASE-30100
> Project: HBase
> Issue Type: Improvement
> Components: Client, documentation
> Reporter: Junegunn Choi
> Assignee: Junegunn Choi
> Priority: Minor
> Labels: pull-request-available
>
> The Javadoc for {{Table.batch(List<? extends Row>, Object[])}} states:
> {quote}
> A null in the result array means that the call for that action failed, even
> after retries.
> {quote}
> This has been wrong since HBASE-6295 (0.96.0). On per-action failure, the
> current implementation stores the {{Throwable}} in the slot, not {{null}}. A
> caller checking only for {{null}} will miss every operational failure.
> The "even after retries" wording is also wrong for the only case {{null}}
> still appears: synchronous input validation in {{RawAsyncTableImpl.batch}}
> (e.g., empty {{Put}} or over-sized cell) throws {{IllegalArgumentException}}
> before any RPC, leaving all slots {{null}}.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
