-
1. Re: Fluent interface on Commands
tom.baeyens Dec 13, 2008 4:28 PM (in response to camunda)let's keep the fluent interface updates for jbpm 4. you can refactor the commands in jbpm 4 in the next iteration (jan-feb) if you have the time.
introducing these kind of changes in 3 might cause unecessary frustration with our current userbase. it being more cool is no reason to break compatibility, i think. -
2. Re: Fluent interface on Commands
kukeltje Dec 13, 2008 6:04 PM (in response to camunda)introducing these kind of changes in 3 might cause unecessary frustration with our current userbase. it being more cool is no reason to break compatibility, i think.
You *think*? LOL... normally you are way more sure about things -
3. Re: Fluent interface on Commands
thomas.diesler Dec 15, 2008 2:48 AM (in response to camunda)It is funny that this thread pops up right now, because over the weekend I investigated whether possible incompatibilities leaked into the jBPM3 code base.
The extend of these incompatibilities can be seen running this command in target/classes for both jbpm-3.3.0 and jbpm-3.3.1find . -name "*.class" | sort | grep class | sed "s/.class//" | xargs javap -classpath . -protected
I put the result in
https://svn.jboss.org/repos/jbpm/jbpm3/trunk/modules/core/src/main/etc/
Obvious incompatibilities are inBatchSignalCommand ChangeProcessInstanceVersionCommand DeleteProcessdefinitionCommand HibernateHelper
Customer code that accesses API that has changed in this incompatible way would break.
Bernd, is it ok for you to roll back these changes until code freeze 20-Dec? If not, please tell me, so we can do it.
This is a blocking issue for the release.
https://jira.jboss.org/jira/browse/JBPM-1919 -
4. Re: Fluent interface on Commands
thomas.diesler Dec 15, 2008 2:58 AM (in response to camunda)Alejandro,
to change HibernateHelper in a compatible way is probably for you.
To make a change in a compatible way, essentially means that you can only add stuff. Incompatible change in behaviour is almost impossible to detect unless yo have especially designed compatibility tests, which jbpm3 does not have.
Hence special care is needed when it comes to behavioural change. -
5. Re: Fluent interface on Commands
camunda Dec 15, 2008 3:45 AM (in response to camunda)Sure, I will take care of the Commands (since I caused it ;-)).
Since the fluent interface is really a cool thing I would add to the "setXxx()" an "xxx()" method with returns the Command itself... This is a compatible change and I hope it is OK for anybody... -
6. Re: Fluent interface on Commands
tom.baeyens Dec 15, 2008 4:20 AM (in response to camunda)ok for me.
-
7. Re: Fluent interface on Commands
camunda Dec 15, 2008 5:17 AM (in response to camunda)Okay, committed. Thomas, maybe you could double check that I haven't missed anything by checking compatibility again?
Thanks. -
8. Re: Fluent interface on Commands
aguizar Dec 15, 2008 4:24 PM (in response to camunda)I'll restore the missing methods in HibernateHelper, but will not restore the configuration cache. jBPM does not actually use it, so it does nothing but leak memory.
-
9. Re: Fluent interface on Commands
aguizar Dec 15, 2008 4:50 PM (in response to camunda)Disregard the previous comment, I fully reverted HibernateHelper. Instead, I changed the only invocation to HibernateHelper that enabled configuration caching.