Hi
1/ I don't think we can support Runtime Data (RTD) defined only
in K3 aspect.
even the pure K3 (sequential) has its RTD in the ecore (in order to
be able to display them in sirius and in the variable view (in the
K3FSM, the current state reference is in the ecore file)
Actually, when using Melange, Melange is used to produce a dsl and
an ecore that has the RTD
If not using Melange, the RTD must be added manually as if it was
produced by Melange. This must include the annotation named "aspect"
that is used to drive the variable view and the trace storage.
BTW, this annotation name: "aspect" is a legacy and has probably a
wrong name since this is not the fact that is comes from an aspect
(ie. produced thanks to Melange) that is important but the fact that
it represents runtime data /dynamic information not matter Melange
was used or not to produce it.
We should have a discussion about a better name for it and open an
issue about it
2/ reminder (if needed), the possibility to set back the full state
works only for the concurrent engine which doesn't rely on the java
stack and thus is really able to reset its full state. The
sequential java engine relies on the java stack, when going backward
we reset the model state but not the java stack state, so we cannot
really run-again it we simply reset the model state, when reaching
the last really executed step, the engine continue normally
Didier
Le 28/08/2018 à 10:07, Julien DeAntoni
a écrit :
Hi,
the multibranch timeline is able to store/restore states of the
model to explore different futures.
It was working ok when used with melange or when the RTD are
directly defined in the ecore file since it was based on
EMFcompare.
Now, we are using pure K3 and the DSL file so that the timeline
save/restore mechanism is broken. Is there already a piece of code
that save/restore the RTD defined in the K3 aspects ?
thanks
j
--
Didier Vojtisek
SED Rennes - DiverSE Team
Univ Rennes, Inria, CNRS, IRISA
Campus de beaulieu
35042 Rennes
02 99 84 75 07
|