[
https://issues.apache.org/jira/browse/HDFS-14483?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16874426#comment-16874426
]
stack commented on HDFS-14483:
------------------------------
Test failures seem related, no?
nit: Was expecting ByteBufferPositionedReadable to be a sub-interface of
ByteBufferReadable/PositionedReadable. Probably fine as is but mildly
'surprising'.
Comparing the BB-based decrypt to the byte[] version, this confuses me:
buf.limit(start + len + Math.min(n - len, localInBuffer.remaining()));
In the byte[] version, its len - n vs n - len figuring how much to decrypt. I
see the byte[] decrypt loop is n < length vs the bb decrypt which is len < n...
so I think its fine but take a look please.
Should these resets be in the finally block just below?
425 buf.position(start + len);
426 buf.limit(limit);
Nice javadoc in ByteBufferPositionedReadable
Appreciate the improvement in testPositionedRead
nit: more permutations in testPositionedReadWithByteBuffer would be
nice-to-have -- though the testByteBufferPread addition is good. Maybe a
follow-on? Could test more edge cases... a failed read or a read that does not
fill the read request amount? The positionedReadCheckWithByteBuffer is nice.
Nice addition of the strncmp check in the test_libhdfs_ops.c file and bulking
up of the pread tests.
Yeah, skip the re-formatting of unrelated code (especially when adds mistake as
in '916 method = "open";')... which adds an offset.
Nice comments added to c function names.
Patch looks great.
> Backport HDFS-3246 ByteBuffer pread interface to branch-2.8.x
> -------------------------------------------------------------
>
> Key: HDFS-14483
> URL: https://issues.apache.org/jira/browse/HDFS-14483
> Project: Hadoop HDFS
> Issue Type: Task
> Reporter: Zheng Hu
> Assignee: Lisheng Sun
> Priority: Major
> Attachments: HDFS-14483.branch-2.8.v1.patch
>
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
