[
https://issues.apache.org/jira/browse/CASSANDRA-9500?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14697307#comment-14697307
]
Benedict commented on CASSANDRA-9500:
-------------------------------------
bq. Permit misaligned flushes is just as inaccurate since now it is incorrectly
named for the constraint that you can't write direct to a channel.
How about {{strictFlushing}} or something? it's confusing reading the trickier
code and seeing a condition on an unrelated concept.
bq. Which code path are you looking for coverage of?
Just a little concerned these were missed by testing. Of course, the endian bug
is hard to trigger without a different endian machine.
Random buffer size would just exercise the slow paths randomly without needing
to change the rest of the tests, but if you're comfortable with the new
coverage that's fine by me
> SequentialWriter should extend BufferedDataOutputStreamPlus
> -----------------------------------------------------------
>
> Key: CASSANDRA-9500
> URL: https://issues.apache.org/jira/browse/CASSANDRA-9500
> Project: Cassandra
> Issue Type: Improvement
> Reporter: Benedict
> Assignee: Ariel Weisberg
> Priority: Minor
>
> SequentialWriter is the last piece left not implementing
> DataOutputStreamPlus. It should not be too tricky to extend it, and it should
> result in improved write throughput. Further, it will permit it to exploit
> the more efficient implementation of writeVInt being delivered in
> CASSANDRA-9499
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
