Using Petri nets for rule termination analysis
Detlef Zimmer, Axel Meckenstock, Rainer Unland · 1997
In Active Database Systems the problem of infinite rule processing may occur.Analysis methods provide an identification of rule sets which canbe guaranteed to terminate or which may lead to an infinite rule processing.We propose Petri nets as the base for a compile-time rule analysis.Pelri nets provide a promising formalism to model and analyse the behaviour of roles and the interactions between them very precisely.Not only rule semantif.8conb'idered by tmmlnation analysis techniques presented in the literalme can easily be modelled but also other parts of the role model as, e.g., complex events.Thus we are able to enhance the results of the t~'minalion analysis methods known to us.In this paper we focus on the analysis of the Petri nets.do not consider complex events and do not allow that at a point in time rules may be triggered more man once and may be executed as often as being triggered.For the detection of complex events Petri nets or higher level Petri nets like Coloured Peui nets have been shown to be suitable [Gat94, ZMU95].Petri nets are used to model possible rule execution traces which are analysed for termination.The information about the interactious between the rules may be token directly from the rule definitions or fi'om the results of other rulc teyminRtiOll analysis techniques, as e.g.[BW94,VS94].As the Petri nets can be generated antomatic~lly there will be no additional specification overhead for the rule programmer (for further devils see [ZMU96]).