topic Historic tracking of multiple end events reached in Alfresco Archive

Historic tracking of multiple end events reached

rgielen — Wed, 02 May 2012 09:22:43 GMT

Hi,as far as I'm concerned there is a problem in the historic data model that keeps us from tracking historic process state correctly. Assumed that following process definition is completely legal in BPMN 2.0 (and we do have similar patterns in our process definitions):[img]https://img.skitch.com/20

Re: Historic tracking of multiple end events reached

trademak — Wed, 02 May 2012 09:43:01 GMT

Hi,

I think your request is valid and it's missing from the current history implementation right now.
Can you create a JIRA?

Thanks,

Re: Historic tracking of multiple end events reached

rgielen — Wed, 02 May 2012 10:24:35 GMT

Done:
http://jira.codehaus.org/browse/ACT-1202

Because of determinism concern filed as bug rather than improvement or feature request.

Thanks,
René

Re: Historic tracking of multiple end events reached

sebastian_s — Wed, 02 May 2012 11:09:14 GMT

From a BPMN 2.0 point of view René's request is valid since the BPMN 2.0 clearly states that a process can have several end events. I would also like to mention that tracking several different end events was possible in jBPM 3.2.

Re: Historic tracking of multiple end events reached

heymjo — Fri, 04 May 2012 05:42:12 GMT

Have you tried fiddling with the HistoryParseListener, history tracking is mostly constructed there i believe

https://fisheye.codehaus.org/browse/~br=root%3A/activiti/activiti/trunk/modules/activiti-engine/src/main/java/org/activiti/engine/impl/history/handler/HistoryParseListener.java?r=3372

Re: Historic tracking of multiple end events reached

bernd_ruecker — Fri, 20 Jul 2012 08:46:58 GMT

We might now tackle a couple of missing history features / flaws for http://www.camunda.com/fox/ 6.1, this could make it into the list 😉

I agree that it is a bug.
My proposal is to add the events correctly to the ACT_HI_ACTINST, which would allow to capture all events and all end events.

But to be correct we would have to remove the PROCINST.END_ACT_ID_ which might cause a lot of users trouble, since they do not use parallelism and maybe are not even aware of the problem. So that would break backwards compatibility. But we could add a warning in the API method for the field and declare it deprecated.

WDYT?

Cheers
Bernd