Messages

1

Bugs and Issues / ESTM: completion transitions from composite states

« on: January 30, 2024, 07:53:18 pm »

Hi all,
we're using Executable State Machines (ESTM) for simulating State Machines STMs. Right now, we're reengineering our control system which is modeled by means of STM. The original code of the controller is working cyclically in opposite of event driven. Thus, all the transitions of the STMs are given as logical conditions, not as triggers. To consider this in STM-model one have to use guards to define completion transitions. Completion transitions utilize completion events, which are generated internally from automatic code generator of ESTM. Because of the fact that completion events are generated only once after the last action has had done and nothing has to be done in this situation, we add unconditioned loop transitions to each state in the model. This causes the for each state the cylic generation of completion events and so the (completion-) transitions will be evaluated this way too. This work very well with normal states, but it don't with composite states. The problem here is that an unconditioned  loop transition on a composite state will not cause genarating cyclic completion events because it never fires. That means that group transitions directly from the composite state to an state outside won't work.
Have anyone any idea to solve the problem ?
Thanks in advance for your help.

BR NB_SMO

2

Bugs and Issues / Re: Executable State Machines

« on: July 27, 2023, 09:39:51 pm »

... ... Meanwhile I've found a solution by myself.
This is based on using a respective unconditioned loop transition ( maintaining a state) for each state. This way, the code generator will caused to insert a so called "completion" event, which will be generated periodically if no other transition is firering. Based on this mechanism, the DO-behaivoir of the states will executed periodically.   There are no death-locks by ommitted evaluations of any expressions.

3

Bugs and Issues / Executable State Machines

« on: July 25, 2023, 08:12:20 pm »

Dear all,

we are using Execuatble State Machines (ESTM) to simulate our STM models. In doing so, we want to control the STMs rather by variables than by events. However, our investigations about this topic has eventually revealed, that all expressions, like transitions, behaviors, effects, which not consist of events, only evaluated once per state configuration. That means, that a transition, which don't fireing at it's evaluation, will never fiering later, even if its expression has menwhile changed, becaus the resp. expression will never be evaluated again in this state configuration.
Has somebody any idea to solve this problem without sacrifice the idea of variable based STM control?
Many thanks in advance.
Best Regards

NB_SMO