-
1. Re: Status update
jmesnil Jan 19, 2010 12:16 PM (in response to timfox)2010/01/19:
- found the leak of https://jira.jboss.org/jira/browse/HORNETQ-255 but did not investigate a fix
- worked on https://jira.jboss.org/jira/browse/HORNETQ-129
to make a long story short, we need to separate the transport from the protocol from the server
so that we can add other protocols without change to the transport or the server
-
2. Re: Status update
jmesnil Jan 20, 2010 12:42 PM (in response to jmesnil)2010/01/20:
- worked on https://jira.jboss.org/jira/browse/HORNETQ-129
I've got code working for sending a Stomp message (CONNECT + SEND + DISCONNECT)
I've also bits of code working to receive a message but the code is missing to send messages to the client
encoded according to the protocol.
I should have this working by tomorrow. I'll maybe sync with the trunk before to take advantage of the
ServerSession cleanup (my code is hacky at this place).
Once I have the stomp receive use case, I'll clean up the code and extract a SPI that needs to be
implemented to support a protocol.
-
3. Re: Status update
jmesnil Jan 21, 2010 1:52 PM (in response to jmesnil)2010/01/21:
- worked on https://jira.jboss.org/jira/browse/HORNETQ-129
updated the branch with tim's refactoring
wasted a few hours fighting with svn branches
Stomp clients can now also receive messages from HornetQ
I did some additional refactoring to clean up the code related to Stomp. Things are straightforwared now:
- if the acceptor is configured to use "stomp", in NettyAcceptor, create delimiter/codec and a handler with the correct
callbacks. The Stomp code is now self-contained in org.hornetq.integration.protocol.stomp
I now need to implement the protocol fully and correctly.
I'd prefer to merge my branch now since I refactored on top of tim's refactoring and continue the protocol
implementation from the trunk. Let's discuss that tomorrow.
-
4. Re: Status update
timfox Jan 21, 2010 3:01 PM (in response to jmesnil)My status
I have almost finished refactoring the code to separate out the concept of protocol fully.
This is almost complete after my next commit. It turned out to be more complex than I originally thought due to references to Packet, RemotingConnection, Channel all over the place.
Now we have a protocol SPI defined. The implementor simply needs to implement ProtocolManager and RemotingConnection.
To illustrate how this is done, I have implemented a simple pointless protocol called AARDVARK - there is an integration test which opens a socket to the server and speaks Aardvark.
In Aardvark, you send the string "AARDVARK!", the server then takes this, then sends a message to a queue with the text "GIRAFFE". This is then consumed and the body of the message sent back to the client.
This should now be ready for Jeff integrating the STOMP protocol in there.
A few things need tidying up, also CORE api is now hacked into RemotingServiceImpl - we need to add a param on the acceptor so the protocol can be configured there.
Also need to deal with Interceptor which currently deals with Packet - this will need to be changed to make it protocol independent.
-
5. Re: Status update
clebert.suconic Jan 21, 2010 7:29 PM (in response to timfox)Imlementing the abstraction at the AS6 testsuite to deploy / undeploy destinations. making good progress.
doing it a way I can back port it.
-
6. Re: Status update
jmesnil Jan 22, 2010 1:45 PM (in response to jmesnil)2010/01/22:
- worked on https://jira.jboss.org/jira/browse/HORNETQ-129
merged the branch in the trunk
-
7. Re: Status update
jmesnil Jan 26, 2010 12:33 PM (in response to jmesnil)2010/01/22:
- worked on https://jira.jboss.org/jira/browse/HORNETQ-129
finished the protocol implementation.
I now need to document it, its limitations and server implementation details.
I'll also write a stomp example (using a TCP client to send/receive stomp messages)
-
8. Re: Status update
ataylor Jan 26, 2010 2:59 PM (in response to jmesnil)Yesterday I was looking at doing the AS JON/JOPR integraton the same way as we do in JBM 1.4. This is not feasible howevere since the management view is build from deployed services not mbeans. i.e. from a deployment file.
today i have looked at implementing a jopr plugin for hornetq. This is the same way that infinispan does and it requires writing a set of component and discovery objects that create the management view. I already have something running locally by the end of tomorrow i should have a basic prototype working in Cleberts branch. Whether this is the correct approach i am not sure. There seems to be several opinions and several implementations for all of this stuff but if i get something done by tomorrow people can take a look and comment.
-
9. Re: Status update
clebert.suconic Jan 26, 2010 7:58 PM (in response to ataylor)Changing and fixing tests and configuration on the AS6 branch. Have found two issues that will need to be investigated tomorrow. (temporary destinations and null clientID... I have opened forum threads for both). -
10. Re: Status update
timfox Jan 26, 2010 8:04 PM (in response to clebert.suconic)My status update:
Mainly doing performance tests today.
Getting some *very* nice results.
-
11. Re: Status update
clebert.suconic Jan 27, 2010 10:29 PM (in response to timfox)I have lots of tests running now.. (with some exceptions, as I posted on the forums).
Tomorrow I will make a few adjustments on security, fix JORAM tests and start on the satelite tests (other tests that are using JMS by "accident" and don't have the real intent to test the JMS itself, such as EJB3, WebServices.. etc).
-
12. Re: Status update
jmesnil Jan 28, 2010 11:59 AM (in response to clebert.suconic)2010/01/19:
- worked on https://jira.jboss.org/jira/browse/HORNETQ-129
added example
fixed tests (making sure command frames are handled by the server using RECEIPT)
working on doc
the protocol is very slow (even for a text-oriented one). I need to profile to see where is the bottleneck
-
13. Re: Status update
clebert.suconic Jan 28, 2010 6:44 PM (in response to timfox)Jan - 28
Making the abstraction on creating tests generic.. applying to other tests now...
On the process, I have a found a few tests written by Adrian that are tightly coupled with the JMX Names. I will be converting those tomorrow.
-
14. Re: Status update
ataylor Jan 29, 2010 4:26 AM (in response to clebert.suconic)28/10
Got the prototype working and cheecked in for a HornetQ jopr plugin. Clebert if u get time could u take a look and comment, log into the admin console and u will see what i have done. I will update the integration thread withe xactly what i have done.
gonna take a look ata users MDB issue as there may be a bug.