Mailing-List: contact commits-help@cassandra.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@cassandra.apache.org
Date: Mon, 2 Mar 2015 09:46:05 +0000 (UTC)
From: "Marcus Eriksson (JIRA)" <jira@apache.org>
To: commits@cassandra.apache.org
Message-ID: <JIRA.12772826.1423171039000.50587.1425289565144@Atlassian.JIRA>
In-Reply-To: <JIRA.12772826.1423171039000@Atlassian.JIRA>
References: <JIRA.12772826.1423171039000@Atlassian.JIRA>
 <JIRA.12772826.1423171039349@arcas>
Subject: [jira] [Commented] (CASSANDRA-8750) Ensure SSTableReader.last
 corresponds exactly with the file end
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/CASSANDRA-8750?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14342982#comment-14342982 ] 

Marcus Eriksson commented on CASSANDRA-8750:
--------------------------------------------

Could you explain this?
{code}
         // buffer offset is always aligned
         bufferOffset = current & ~(buffer.length - 1);
+        if (bufferOffset + validBufferBytes > length())
+            validBufferBytes = (int)(length() - bufferOffset);
{code}
are we not returning the valid byte count out of metadata.compressor().uncompress(..) ?

> Ensure SSTableReader.last corresponds exactly with the file end
> ---------------------------------------------------------------
>
>                 Key: CASSANDRA-8750
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-8750
>             Project: Cassandra
>          Issue Type: Improvement
>          Components: Core
>            Reporter: Benedict
>            Assignee: Benedict
>            Priority: Minor
>             Fix For: 2.1.4
>
>
> Following on from CASSANDRA-8744, CASSANDRA-8749 and CASSANDRA-8747, this patch attempts to make the whole opening early of compaction results more robust and with more clearly understood behaviour. The improvements of CASSANDRA-8747 permit is to easily align the last key with a summary boundary, and an index and data file end position. This patch modifies SegmentedFile to permit the provision of an explicit length, which is then provided to any readers, which enforce it, ensuring no code may accidentally see an end inconsistent with the one advertised. 


--
This message was sent by Atlassian JIRA
(v6.3.4#6332)