Thanks Christophe. I will make the change to see it work; but based on your comments, I'll probably switch to one of the other front-ends.
Thanks,
Steve
What I mean is that eUML is not used as long as other front-ends and therefore less tested. But there has been so far no bug I haven't been able to fix within a few days. In the case you reported, I have a fix which seems to be working (and reduces the syntax needed), I just need to play with it a bit more before committing it.
Hi, as promised, I fixed this problem (trunk rev. 76655). It even requires less typing. You define a submachine with front-end like always and use it in another state machine. In your case, you'd have: typedef boost::msm::back::state_machine<Normal_T> Normal_TImpl; The remaining code is unchanged: Alarm_TImpl alarm; BOOST_MSM_EUML_DECLARE_TRANSITION_TABLE(( normal + evAlarm / resetDeviceInited() == alarm, alarm + evAlarmCleared == normal ),transition_table) You probably moved to another front-end, but in case you want to try eUML, this problem is fixed. Cheers, Christophe