You are viewing a plain text version of this content. The canonical link for it is here.

Posted to commits@cassandra.apache.org by "Benedict (JIRA)" <ji...@apache.org> on 2015/08/13 12:55:45 UTC

[jira] [Commented] (CASSANDRA-10064) Serialization should avoid calculating serializedSize up-front

    [ https://issues.apache.org/jira/browse/CASSANDRA-10064?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14695056#comment-14695056 ] 

Benedict commented on CASSANDRA-10064:
--------------------------------------

There is certainly an overlap with CASSANDRA-8100, but we use serializedSize inside of serialization, not just for intra cluster message payload size. It may be that we only need CASSANDRA-8100 and a bit of refactoring, however this approach would give us a general purpose utility for ensuring we never need a serializedSize for easy serialization.

> Serialization should avoid calculating serializedSize up-front
> --------------------------------------------------------------
>
>                 Key: CASSANDRA-10064
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-10064
>             Project: Cassandra
>          Issue Type: Improvement
>          Components: Core
>            Reporter: Benedict
>             Fix For: 3.x
>
>
> There are multiple approaches that could be used here, but the simplest would be to extend DataInputPlus and DataOutputPlus to support marking the start/end of a variable length (and unkown upfront) encoding. By itself this only buys us more efficient encoding, however it opens the door to the sending of results before they have been fully materialized (but this has many other challenges).



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