On Jan 9, 2013, at 10:43 AM, Markus Klotzbuecher <markus [dot] klotzbuecher [..] ...> wrote:

> Hi Dustin,
>
> On Wed, Jan 09, 2013 at 10:23:07AM -0600, Gooding, Dustin R. (JSC-ER411) wrote:
>> I started working with rFSM and quickly found something unexpected. I don't
>> know why one of these machines works and the other doesn't. I'd appreciate
>> some guidance.
>>
>> Essentially, the following machines are identical, it's just that one has an
>> extra state and some extra transitions. The idea is that there's two primary
>> states, and inside one of those states are a handful of sub-states. Those
>> substates are configured (essentially) where there's one initial sub-state that
>> can transition to any of the others. And the others can transition from one to
>> the next (in one direction). There is also an "error" state (the "last" state)
>> that all sub-states can transition to, and it can transition back to the
>> initial sub-state. With 5 sub-states, it works. With 6, I get errors.
>>
>> The state machine(s), I switch back and forth by selectively commenting the "5"
>> or "6" section:
>> /**
>> return rfsm.state
>> {
>> sA = rfsm.state{ entry=function() print("sA") end },
>>
>> sB = rfsm.state{ entry=function() print("sB") end,
>>
>> initial = rfsm.conn{},
>> rfsm.trans {src="initial", tgt="sB_s1"},
>>
>> sB_s1 = rfsm.state{ entry=function() print("sB_s1") end },
>> sB_s2 = rfsm.state{ entry=function() print("sB_s2") end },
>> sB_s3 = rfsm.state{ entry=function() print("sB_s3") end },
>> sB_s4 = rfsm.state{ entry=function() print("sB_s4") end },
>> sB_s5 = rfsm.state{ entry=function() print("sB_s5") end },
>>
>> -- 5 states
>> rfsm.trans {src="sB_s1", tgt="sB_s2", events={'e_sB_s2'}},
>> rfsm.trans {src="sB_s1", tgt="sB_s3", events={'e_sB_s3'}},
>> rfsm.trans {src="sB_s1", tgt="sB_s4", events={'e_sB_s4'}},
>>
>> rfsm.trans {src="sB_s2", tgt="sB_s3", events={'e_sB_s3'}},
>> rfsm.trans {src="sB_s3", tgt="sB_s4", events={'e_sB_s4'}},
>>
>> rfsm.trans {src="sB_s1", tgt="sB_s5", events={'e_sB_s5'}},
>> rfsm.trans {src="sB_s2", tgt="sB_s5", events={'e_sB_s5'}},
>> rfsm.trans {src="sB_s3", tgt="sB_s5", events={'e_sB_s5'}},
>> rfsm.trans {src="sB_s4", tgt="sB_s5", events={'e_sB_s5'}},
>> rfsm.trans {src="sB_s5", tgt="sB_s1", events={'e_sB_s1'}},
>>
>> -- 6 states
>> -- sB_s6 = rfsm.state{ entry=function() print("sB_s6") end },
>>
>> -- rfsm.trans {src="sB_s1", tgt="sB_s2", events={'e_sB_s2'}},
>> -- rfsm.trans {src="sB_s1", tgt="sB_s3", events={'e_sB_s3'}},
>> -- rfsm.trans {src="sB_s1", tgt="sB_s4", events={'e_sB_s4'}},
>> -- rfsm.trans {src="sB_s1", tgt="sB_s5", events={'e_sB_s5'}},
>>
>> -- rfsm.trans {src="sB_s2", tgt="sB_s3", events={'e_sB_s3'}},
>> -- rfsm.trans {src="sB_s3", tgt="sB_s4", events={'e_sB_s4'}},
>> -- rfsm.trans {src="sB_s4", tgt="sB_s5", events={'e_sB_s5'}},
>>
>> -- rfsm.trans {src="sB_s1", tgt="sB_s6", events={'e_sB_s6'}},
>> -- rfsm.trans {src="sB_s2", tgt="sB_s6", events={'e_sB_s6'}},
>> -- rfsm.trans {src="sB_s3", tgt="sB_s6", events={'e_sB_s6'}},
>> -- rfsm.trans {src="sB_s4", tgt="sB_s6", events={'e_sB_s6'}},
>> -- rfsm.trans {src="sB_s5", tgt="sB_s6", events={'e_sB_s6'}},
>> -- rfsm.trans {src="sB_s6", tgt="sB_s1", events={'e_sB_s1'}},
>> },
>>
>> initial = rfsm.conn{},
>> rfsm.trans {src="initial", tgt="sA"},
>>
>> rfsm.trans {src="sA", tgt="sB", events={'e_sB'}},
>> rfsm.trans {src="sB", tgt="sA", events={'e_sA'}},
>> }
>> **/
>>
>>
>>
>>
>> When I run rfsm-sim, I get the following output
>>
>> /** with 5 states
>> $ rosrun rfsm rfsm-sim test.lua
>> Lua 5.1.4 Copyright (C) 1994-2008 Lua.org, PUC-Rio
>> rFSM simulator v0.1, type 'help()' to list available commands
>>> pp()
>> root
>> sA[LS]
>> sB[CS]
>> sB_s5[LS]
>> sB_s3[LS]
>> sB_s4[LS]
>> sB_s2[LS]
>> sB_s1[LS]
>>
>>> run()
>> sA
>> active: root.sA(done)
>> queue:
>>>
>> **/
>>
>> /** with 6 states
>> $ rosrun rfsm rfsm-sim test.lua
>> Lua 5.1.4 Copyright (C) 1994-2008 Lua.org, PUC-Rio
>> rFSM simulator v0.1, type 'help()' to list available commands
>>> pp()
>> root
>> sA[LS]
>> sB[CS]
>> sB_s5[LS]
>> sB_s3[LS]
>> sB_s4[LS]
>> sB_s2[LS]
>> sB_s6[LS]
>> sB_s1[LS]
>>
>>> run()
>> sA
>> /opt/orocos/rfsm/tools/rfsm-sim: line 10: 14520 Aborted (core
>> dumped) env $LUA -i $SCRIPTPATH/rfsm-sim.lua $1
>> **/
>
> Fortunately rFSM can't really crash by themself, so it means it's
> somebody else fault :-)
>
> My guess it is graphiz, which for unknown reasons sometimes crashes
> during layout. Although you may not be using it directly, the figures
> are still generated using rfsm-sim (probably we should not do that).
>
> Could you try if the problem disappears with the attached patch
> against rfsm-sim?
>
>> What's weird is that when I run these through rfsm-viz to generate UML
>> diagrams, they both work fine.
>>
>> Why does 5 work and 6 fail? Why, in the 6 state one, is sB_s6 so low in the
>> pp() list?
>
> The list is unordered and not sorted in any way.
>
> Markus
>
>
>
> diff --git a/tools/rfsm-sim.lua b/tools/rfsm-sim.lua
> index ea447d8..4baba4c 100644