LargeMessage deliverer - https://jira.jboss.org/jira/browse

[clebert.suconic]

A simple fix, would be to create a method called cloneForDeliver on the LargeMessage interface, and use it on the ServerConsumer::LargeMessageDeliverer. This way we could reuse the same buffer on NullPersistence, and create a new file on Journal.

[timfox]

What do you mean "create a new file on Journal" ?

[clebert.suconic]

a new File Handler. reuse the same file though.

[timfox]

I don't see how this is related to the journal.

[clebert.suconic]

There are two implementations to LargeMessage:


http://anonsvn.jboss.org/repos/hornetq/trunk/src/main/org/hornetq/core/persistence/impl/nullpm/NullStorageLargeServerMessage.java

http://anonsvn.jboss.org/repos/hornetq/trunk/src/main/org/hornetq/core/persistence/impl/journal/JournalLargeServerMessage.java


When I said on Null Persistence, and on Journal, I was referring to those implementations individually.


the JournalLargeServerMessage is just the LargeMessage used when using the Journal. In fact it just uses a file.

[ataylor]

A simple fix, would be to create a method called cloneForDeliver on the LargeMessage interface, and use it on the ServerConsumer::LargeMessageDeliverer. This way we could reuse the same buffer on NullPersistence, and create a new file on Journal.

Thats similar to what i am doing. when large messages are encoded i pass round an encoding context, this context takes care of making a copy of SequentialFile (i.e. with a different file handle) and the large message encode method encodes using the context. Doing it this way i may be able to get rid of separate encode methods for normal messages and large messages, simplifying things.