April 2009 - jboss-dev-forums - Jboss List Archives

[Design of POJO Server] - Removing ClassLoading from jboss-bootstrap

by ALRubinger

https://jira.jboss.org/jira/browse/JBBOOT-24 As it stands, Bootstrap is sensitive to arbitrary JARs on the application CP (and picked up by the extension ClassLoader). This becomes especially apparent when starting up from an Embedded environment, where we may have a full dependency set on the test CP. AS currently sidesteps this problem by careful inclusion of classes in the run.jar. What I'd like to do is remove setting of the TCCL from jboss-bootstrap entirely, leaving this task up to the client (launcher). Some local tests yield much better results and allow the bootstrap to become adaptable to a wider range of runtime environments. Can anyone weigh in on: * Removal of NoAnnotationCL - a URLCL which is commented to be used in RMI. Is this still the case? * AbstractServerImpl.start() to use the TCCL. Currently it's using 2 different CLs (one for starting the ServerImpl, one for the bootstraps). With no objections I'll plug ahead, test, and patch AS to match. S, ALR View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4223119#4223119 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4223119

17 years

1
0
0 / 0

[Design of Clustering on JBoss] - Re: JBAS-5703 - Make HA-JNDI use wait for first positive res

by bstansberry＠jboss.com

"galder.zamarreno(a)jboss.com" wrote : Re: https://jira.jboss.org/jira/browse/JBAS-5703 | | I've attached a patch containing fix for this. Few notes: | Great; thanks. anonymous wrote : 1. I've tested this with a 3 node cluster (node1,node2 and node3) and two ejbs (ejbA and ejbB where ejbA calls ejbB via HAJNDI). Deployment wise, ejbA is deployed in node1 and ejbB is deployed in node3. I was able to make a call successfully via ejbA in node1 and was able to check via logs that once response from node3 was received, the lookup returned. Now, my question is, how to create a unit test around this? Not easy... Testing can be done in layers: a) Unit test your LookupSucceededFilter. That's really the key thing here. b) Unit test RspFilterAdapter. c) A test of ClusterPartition.callMethodOnCluster where ResponseFilter param is passed. One way to do that at least partially is to have a service that calls method public boolean echo(boolean arg), where response is whatever arg was. Call with a filter that only accepts "true". Invoke method twice, passing "true" then "false"; should get a response to first call, none to second call. d) A full test of the HA-JNDI functionality. That can wait until we make ClusterPartition more of a POJO, something where a couple can be instantiated in a unit test the way JBC can. You can file a JIRA to add such a test. anonymous wrote : 2. I noted this as a todo in the patch: ResponseFilter/RspFilter API: If needMoreResponses() took response and sender as well, class implementations could be Immutable and same instance could be passed to all lookups. As it is right now, they can't and new instances need to be generated for every lookup. Having 2 methods lets you filter responses out of the response list while still accepting more responses. To get the same thing out of one method you'd have to change the return type to something beyond simple boolean, perhaps Boolean but better an enum. This would be a JGroups API change. anonymous wrote : 3. Could response filtering done in ClusterPartition.processResponseList() be moved here? Any NoHandlerForRPC would be caught here immediately and could potentially avoid double processing. Thoughts? Something to think about; could be done in RspFilterAdapter. Hmm -- don't think it works well for the ordinary case where you want responses from everyone. To return properly from needMoreResponses() the RspFilterAdapter would have to know how many responses it is expecting and count how many it has gotten. And deal with suspicions/view changes that may occur in the meantime. View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4223115#4223115 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4223115

17 years

1
0
0 / 0

[Design of Messaging on JBoss (Messaging/JBoss)] - jbm-jms schema...

by clebert.suconic＠jboss.com

Is that ok if I change jbm-jms.xsd to make connection-factory minOccurs="0"? I'm defining my own jbm-jms.xml on paging-examples, and I don't need any ConnectionFactories on that example, and the parsing would fail if I don't make that change. View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4223113#4223113 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4223113

17 years

1
0
0 / 0

[Design of Messaging on JBoss (Messaging/JBoss)] - Re: QueueDeployer?

by clebert.suconic＠jboss.com

I have figured out. But I still don't understand is: - Why we're not using jbm-jms.xml to deploy the examples? - Have we thought about using masks on the deployers, such as jbm-jms*.xml? In a way the user could have its own config files? View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4223112#4223112 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4223112

17 years

1
0
0 / 0

[Design of Clustering on JBoss] - JBAS-5703 - Make HA-JNDI use wait for first positive respons

by galder.zamarreno＠jboss.com

Re: https://jira.jboss.org/jira/browse/JBAS-5703 I've attached a patch containing fix for this. Few notes: 1. I've tested this with a 3 node cluster (node1,node2 and node3) and two ejbs (ejbA and ejbB where ejbA calls ejbB via HAJNDI). Deployment wise, ejbA is deployed in node1 and ejbB is deployed in node3. I was able to make a call successfully via ejbA in node1 and was able to check via logs that once response from node3 was received, the lookup returned. Now, my question is, how to create a unit test around this? Not easy... 2. I noted this as a todo in the patch: ResponseFilter/RspFilter API: If needMoreResponses() took response and sender as well, class implementations could be Immutable and same instance could be passed to all lookups. As it is right now, they can't and new instances need to be generated for every lookup. 3. Could response filtering done in ClusterPartition.processResponseList() be moved here? Any NoHandlerForRPC would be caught here immediately and could potentially avoid double processing. Thoughts? View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4223109#4223109 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4223109

17 years

1
0
0 / 0

[Design of Management Features on JBoss] - Re: JBAS-6672, null values and defaults

by charles.crouch＠jboss.com

"scott.stark(a)jboss.org" wrote : | That a removed property has no value applied to the metadata and the metadata simply defaults to a value will apply to all managed property types, including primitives. Is that the function you are questioning? | Yep, if we can get that, I think we're good for now. View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4223099#4223099 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4223099

17 years

1
0
0 / 0

[Design of Messaging on JBoss (Messaging/JBoss)] - QueueDeployer?

by clebert.suconic＠jboss.com

Why we are not using the QueueDeployer in our examples? For instance.. I'm having to specify a specific AddressSetting for PagedDestination, but the deployment of the queue is being done through JMSExample::deployQueues. I looked through QueueDeployer code, and it seems inactive. Maybe I'm missing something there. View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4223097#4223097 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4223097

17 years

1
0
0 / 0

[Design of JBoss Wiki] - JBoss Wiki pages updated on JBoss Community Wiki

by piergiorgiolucidi

I have updated JBoss Wiki pages on JBoss Community Wiki, you can find all the updates from main page here: http://www.jboss.org/community/wiki/JBossWiki I'll commit on JBoss Wiki project a directory called infoDeploy to describe Maven installation and building. Hope this helps. View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4223093#4223093 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4223093

17 years

1
0
0 / 0

[Design of Management Features on JBoss] - Re: JBAS-6672, null values and defaults

by scott.stark＠jboss.org

"charles.crouch(a)jboss.com" wrote : | So overall the most important thing is to have property.setRemoved() cause nothing to be persisted for the managed property, and that a default (if specified somewhere) is applied to the actual running instance. [This could just be letting the component itself choose the default, I don't think it has to come from the ManagedProperty annotation for CR1, though it would make sense in the future.] This is what I see as the core issue that needs resolving for CR1. | | If we don't have this function for managedproperty's which are primitives then that's not great, but I don't necessarily see it as a blocker since there probably not that many such properties. That a removed property has no value applied to the metadata and the metadata simply defaults to a value will apply to all managed property types, including primitives. Is that the function you are questioning? View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4223090#4223090 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4223090

17 years

1
0
0 / 0

[Design of JBoss jBPM] - Re: jBPM 4 Email Question

by bradsdavis

>From my experience in the field, here are two features we should actively pursue. 1) Multiple email servers. Reason: a lot of companies have multiple SMTP servers. The reason is, generally one sends internally and one externally. I think we could accomplish this by having a configuration for the email servers in the mail configuration. This would be a list of servers, which could take simple domain filtering. For example servers could include "*jboss.org" and that would indicate the mail server can serve all jboss.org emails. We could also support pop or smtp mail pretty easily. 2) Email Attachments are often an interest for clients. An example would be sending an order in PDF format when the task to sign off on the order is produced. We could create an email producer interface that we could use to make email producers plugable. It could take the environment context as a parameter, and return an email object [see apache email commons]. That would make the interface flexible enough to return either a simple text email, HTML email, or Multipart email with attachments. The email producer that we could ship could target using jsr223 or velocity or freemarker. I have noticed all of those have been tossed around as possible templating. View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4223089#4223089 Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4223089

17 years

1
0
0 / 0

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

jboss-dev-forums April 2009