September 2008 - jbpm-dev - Jboss List Archives

[Design of JBoss jBPM] - Initialise thread context classloader

by camunda

Hi. I am working on https://jira.jboss.org/jira/browse/JBPM-1448 (jBPM does not initialize thread context classloader). First of all, why I want to do this: This enables you to put Interfaces or other stuff used in your action classes in the par, so they end up in the database. Because the ProcessClassLoader will become ContextClassLoader it is used to load everything correctly... But this is only half the way I want to go. Image you put jbpm in your app-server (or ESB) once, but have different applications (ear, esb, ...) using it. Now I write an own small MBean, which is a registry where processes register which class loaders are responsible for "their" action classes. My own ProcessClassLoader now also looks in this registry and by doing so, it can even load action classes or other stuff correctly from the right EAR, even if a process was maybe started by the web-console, not knowing the ear... But more on this idea later. Back to setting the ContextClassLoader. My approach to implement it is adding this code | ClassLoader oldCl = Thread.currentThread().getContextClassLoader(); | try { | Thread.currentThread().setContextClassLoader( | ClassLoaderUtil.getProcessClassLoader(processDefinition)); | | .... existing code ... | } | finally { | Thread.currentThread().setContextClassLoader(oldCl); | } | to the following methods: - Action.execute - Decision.execute - TaskController.initializeVariables - TaskController.submitParameters - AssignmentHandler.performAssignmentDelegation On all these places I have the ProcessDefinition on hand, so it is possible to do... These are all places I found where the ProcessClassLoader is used and all Delegation use cases I remembered. So it is not too much work to do and consistent overall in jBPM (not like the proposed workaround to build an own base action class where the context class loader is set). What do you guys think? Cheers Bernd View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4178720#4178720 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4178720

15 years, 8 months

1
0
0 / 0

[Design of JBoss jBPM] - Re: gwt-console in distribution

by heiko.braun＠jboss.com

Yes, the gwt:compile take some time. But it's only required for web mode. So if you want to run the console in hosted mode a simple 'mvn compile gwt:gwt' is sufficient. This reduces the turnaround time a lot. Add a "mvn -o" to it and becomes a pleasure to work with. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4178716#4178716 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4178716

15 years, 8 months

1
0
0 / 0

[Design of JBoss jBPM] - Re: gwt-console in distribution

by thomas.diesler＠jboss.com

Thanks, this dropped the build time by 40% http://jbpm.dyndns.org:8280/hudson/job/jBPM-Matrix/buildTimeTrend View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4178665#4178665 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4178665

15 years, 8 months

1
0
0 / 0

Re: jBPM SEAM integration

by Tom Baeyens

let's say friday around 11am. by that time, i'm sure that i will have time be fully prepared. regards, tom. Pete Muir wrote: > > On 24 Sep 2008, at 14:44, Tom Baeyens wrote: > >> Pete, >> >> There are 2 topics I would like to discuss with you: >> >> 1) https://jira.jboss.org/jira/browse/JBPM-1166 > > Ok. I thought we had sorted this? > >> 2) Moving Pageflow to SEAM. The problem of having pageflow in jBPM is >> that that introduces a circular dependency. The pageflow node depends >> on SEAM and SEAM depends on jBPM pageflow. This circular dependency >> also prevents us from testing pageflow in our own codebase. > > What does this involve? The public class Page extends Node is already in > the seam code base, so is this just some configuration? > >> Next monday will be JIRA freeze we'll have to decide by then how we >> are going to deal with this situation. >> >> Are you available for a call about this topic later this week ? > > Perhaps sometime on Friday morning? -- regards, tom.

15 years, 8 months

1
0
0 / 0

jBPM SEAM integration

by Tom Baeyens

Pete, There are 2 topics I would like to discuss with you: 1) https://jira.jboss.org/jira/browse/JBPM-1166 2) Moving Pageflow to SEAM. The problem of having pageflow in jBPM is that that introduces a circular dependency. The pageflow node depends on SEAM and SEAM depends on jBPM pageflow. This circular dependency also prevents us from testing pageflow in our own codebase. Next monday will be JIRA freeze we'll have to decide by then how we are going to deal with this situation. Are you available for a call about this topic later this week ? -- regards, tom.

15 years, 8 months

1
0
0 / 0

[Design of JBoss jBPM] - Re: gwt-console in distribution

by heiko.braun＠jboss.com

Sure, I will take a look at it. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4178486#4178486 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4178486

15 years, 8 months

1
0
0 / 0

[Design of JBoss jBPM] - Re: add

by jbarrez

Tom, I understand the reason for the difference now... damn it's more difficult than I thought :p But does that mean that a Command should be an internal class and that clients shouldn't use it (but the CommandService instead)? View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4178480#4178480 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4178480

15 years, 8 months

1
0
0 / 0

[Design of JBoss jBPM] - Re: add

by tom.baeyens＠jboss.com

there is a difference. the command is executed within a environment (read:transaction). an exception there will rollback. in this case it is a convenience so that you don't have to rethrow checked exceptions. the service methods are exposed to the client. that is why they are different. anyway... that is the reasoning behind it. doesn't mean it's the right thing. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4178474#4178474 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4178474

15 years, 8 months

1
0
0 / 0

[Design of JBoss jBPM] - Re: gwt-console in distribution

by thomas.diesler＠jboss.com

Heiko, would it be possible to build the gwt-console as part of the distro profile i.e. in the same way as the userguide? This would reduce the normal change build test cycle considerably. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4178473#4178473 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4178473

15 years, 8 months

1
0
0 / 0

[Design of JBoss jBPM] - Re: add

by jbarrez

There is certainly a mismatch between the 2 ... Having the throws clause will make it cumbersome if no real Exception handling is done (95% of the cases) ... but it sure is easy when you need it... I'll have to think about it before I can give my final answer :p View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4178420#4178420 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4178420

15 years, 8 months

1
0
0 / 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

jbpm-dev September 2008