[
https://issues.apache.org/jira/browse/CASSANDRA-1661?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12924617#action_12924617
]
Jonathan Ellis commented on CASSANDRA-1661:
-------------------------------------------
bq. simplified with judicious use of mark() and get()
I worry that we're going to introduce very very tricky bugs if we don't treat
position as immutable. If BB/Thrift were sane about using offset to represent
the start of what we're allowed to mess with, that would be another story, but
it's not -- Thrift uses wrap to generate the BB it gives us, so position is the
only indication we have of what a given BB is supposed to cover.
bq. ByteBufferTest is literally a test of ByteBuffers: I don't think it should
be in C*
OTOH ByteBuffer is a tricky enough API (see: this ticket :) that I don't mind
having BBT as an illustration of what is going on if not exactly a test.
Maybe docstring in BBUtil instead of actual tests?
> Use of ByteBuffer limit() must account for arrayOffset()
> ---------------------------------------------------------
>
> Key: CASSANDRA-1661
> URL: https://issues.apache.org/jira/browse/CASSANDRA-1661
> Project: Cassandra
> Issue Type: Bug
> Components: Core
> Affects Versions: 0.7.0
> Reporter: T Jake Luciani
> Assignee: T Jake Luciani
> Fix For: 0.7.0
>
> Attachments: 1661_v1.txt
>
>
> There are a few places in the code where it loops across a byte buffers
> backing array wrong:
> for (int i=bytes.position()+bytes.arrayOffset(); i<bytes.limit(); i++)
> This is incorrect as the limit() does not account for arrayOffset()
> for (int i=bytes.position()+bytes.arrayOffset();
> i<bytes.limit()+bytes.arrayOffset(); i++)
> is the correct code.
> There is also a few places where the unit tests would fail if we used non
> wrapped byte arrays.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
