[
https://issues.apache.org/jira/browse/CASSANDRA-9499?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14597794#comment-14597794
]
Ariel Weisberg commented on CASSANDRA-9499:
-------------------------------------------
I'm confused. Are we doing this as one commit or two? Previously I thought that
adding varint encoding was going to be separate from removing Encoded streams.
You can get github (and git can probably do this as well) to provide you with a
squashed commit without destroying the commit history on the development
branch.
https://github.com/apache/cassandra/compare/trunk...aweisberg:C-9499.diff
Using the temporary buffer isn't safe. You can construct scenarios where the
buffer is retrieved and used twice. Once for varint and once when wrapping a
bytebuffer at which point ensureRemaining() could clobber it. It's a dubious
pattern in the first place. The only way to make it safe IMO is to have a list
of buffers and return them when done.
I could fix it by calling ensure remaining before doing the encoding in BDOSS
but that still leaves me thinking it's data corruption waiting to happen.
> Introduce writeVInt method to DataOutputStreamPlus
> --------------------------------------------------
>
> Key: CASSANDRA-9499
> URL: https://issues.apache.org/jira/browse/CASSANDRA-9499
> Project: Cassandra
> Issue Type: Improvement
> Components: Core
> Reporter: Benedict
> Assignee: Ariel Weisberg
> Priority: Minor
> Fix For: 3.0 beta 1
>
>
> CASSANDRA-8099 really could do with a writeVInt method, for both fixing
> CASSANDRA-9498 but also efficiently encoding timestamp/deletion deltas. It
> should be possible to make an especially efficient implementation against
> BufferedDataOutputStreamPlus.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
