[Design of JBoss jBPM] - Re: Removed Timer/autosave in Revision 4006?
by alex.guizar@jboss.com
"camunda" wrote : In my case, we have an own ExecuteJobCommand, since we do some additional stuff. So I had to change that as well to get the Tests green. Maybe other people do the same stuff?
Yes, the Seam integration also uses a custom job executor and they may rely on the jobs autosaving the process instance.
"camunda" wrote : Couldn't we move this to the JobSession.loadJob() method? Now we have to spread this logic to JobExecutor, ExecuteJobCommand's and so on.
Seems like a good proposal, except that loadJob could also be used under contexts that do not require saving the process instance (e.g. a monitoring console). I have the following two proposals:
1. Introduce a loadJobForUpdate method, possibly in JbpmContext (since no XxxSession has any xxxForUpdate method).
2. Forget about it, and just revert.
I'm leaning towards (1). Any thoughts?
"camunda" wrote : I fly to Mexico next week. Maybe we could just meet for beer on around Tuesday 10.03. in Mexico City?
Crap, what a bad coincidence. Right now I'm in Houston, USA, and will be back to Mexico City by March 14th. Perhaps we could meet on your way back?
View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4213906#4213906
Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4213906
15 years, 10 months
[Design of JBoss jBPM] - Removed Timer/autosave in Revision 4006?
by camunda
Bernd wrote :
| > Hi Alex!
| > With Revision 4006 you removed the fix for
| > https://jira.jboss.org/jira/browse/JBPM-1015 from the Timer class.
| > basically this code:
| > Was this by accident? Or what rational was behind this? Can I
| > re-include it?
| > Because now, Bug 1015 would occur again with jbpm 3.2.6
|
Alejandro wrote :
| While working on JBPM-2036, I noticed that the addAutoSaveToken call was
| duplicated in all job classes, so I moved it to JobExecutorThread. Now
| that you mention it, the code should be added to ExecuteJobCommand as
| well.
|
| Since no test failed, I assumed the net effect was the same. I guess it
| broke something on your side. Care to tell me how to reproduce?
|
Hi Alex.
Couldn't we move this to the JobSession.loadJob() method? Now we have to spread this logic to JobExecutor, ExecuteJobCommand's and so on.
In my case, we have an own ExecuteJobCommand, since we do some additional stuff. So I had to change that as well to get the Tests green. Maybe other people do the same stuff?
And in JobSession it would be nice, since then it is "close" to the code handling persistence, or not?
Remarks?
@Ronald: Yeah, I see a slight chance, that it was missed to write a test for executing jobs via Command ;-)
Cheers
Bernd
P.S @Alex: I fly to Mexico next week. Maybe we could just meet for beer on around Tuesday 10.03. in Mexico City?
View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4213877#4213877
Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&p=4213877
15 years, 10 months