[infinispan-issues] [JBoss JIRA] Commented: (ISPN-1326) Multiplexing marshalling architecture

Galder Zamarreño (JIRA) jira-events at lists.jboss.org
Thu Sep 8 10:43:26 EDT 2011


    [ https://issues.jboss.org/browse/ISPN-1326?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12627268#comment-12627268 ] 

Galder Zamarreño commented on ISPN-1326:
----------------------------------------

This is pretty much done since Tuesday but some issues in the testsuite have halted my progress. I've fixed them all but there's one in particular that it's doing my head in. I'm getting a random test failure which, after some investigation, so far is pointing towards some corruption without inner JBoss Marshalling state from where unmarshallers are retrieved. Given that there's no logging there, I'm falling back on ByteMan to help me here. I'm slowly making progress but I'll get there :). Still haven't figured out the root cause but hope to have a better idea fairly soon.

> Multiplexing marshalling architecture
> -------------------------------------
>
>                 Key: ISPN-1326
>                 URL: https://issues.jboss.org/browse/ISPN-1326
>             Project: Infinispan
>          Issue Type: Enhancement
>          Components: Marshalling
>            Reporter: Galder Zamarreño
>            Assignee: Galder Zamarreño
>             Fix For: 5.1.0.BETA1
>
>
> The current solution for ISPN-1205 involves sending the cache name along with each MarshalledValue marshalled payload which is not optimal. The reason this is needed is to be able to locate the marshaller that the MV needs, which needs to be given on instance creation.
> To make this more efficient, a multiplexing marshalling architecture is needed where the cache name is marshalled way before anything else in the command and based on that, the right marshaller can be picked and passed down to the externalizers to use as needed, i.e. MarshalledValue.
> The reason for doing this separately to ISPN-1205 is because some decent restructuring is needed to get this done cos currently the cache name is embedded within the actual command parameters, which makes it ackward to select a marshaller there and then continue unmarshalling with that marshaller. 

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira

       



More information about the infinispan-issues mailing list