[
https://issues.apache.org/jira/browse/CASSANDRA-2361?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13012119#comment-13012119
]
Sylvain Lebresne commented on CASSANDRA-2361:
---------------------------------------------
This looks ok, but given the Merkle tree can be have a significant size (in
term of nodes), it would maybe be worst it to use a byte to distinguish between
a Inner and Leaf in the serialized form instead of the full class name. I
understand that Java serialization wasn't necessarily very compact, but since
we have a chance to do better.
> AES depends on java serialization
> ---------------------------------
>
> Key: CASSANDRA-2361
> URL: https://issues.apache.org/jira/browse/CASSANDRA-2361
> Project: Cassandra
> Issue Type: Improvement
> Components: Core
> Affects Versions: 0.7.4
> Reporter: Gary Dusbabek
> Assignee: Gary Dusbabek
> Priority: Minor
> Fix For: 0.7.5, 0.8
>
> Attachments: 2361-part1.patch,
> v1-0001-ICompactSerializers-for-MerkleTrees.txt,
> v1-0002-git-rid-of-annoying-AES-emissions.txt
>
>
> 0.8 should be able to run in the same cluster as 0.7. AES uses java
> serialization which means that Token serialization stands a good chance of
> being brittle. This needs to be fixed.
> 1. place a hard-coded serialVersionUID for Token in 0.7.5.
> 2. have AES use ICompactSerializer in place of ObjectInputStream in 0.8.
> This would be a good opportunity to audit the code for imprudent uses of
> Object[Input|Output]Stream and get those cleaned up. Also, it will exercise
> the versioning code a bit.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
