Introduction

GPAL (GM Petri net Abstraction Layer) provides an abstract view on SCPNs. It provides facilities to load a given net from its XML file or to generate a new net from scratch.

Various model checks make sure that the C++ code that a code generator will produce from the net, is most likely to be correct. All expressions that a user may enter (like inscriptions, firing delays, guard functions etc.) can be checked for syntactical and type correctness.

Another important feature of GPAL is a process named inlining of a model. Models can contain instances of modules that may be kept in external libraries. To ease code generation such a net must be inlined which means that all instances of such modules will be 'instantiated' (in the original sence of the word), i.e. a copy of the implementation of the module will be created for each instance and replace it.

An inlined SCPN can be stored to an XML file.

Grammars

As mentioned above GPAL can perform syntactical checks on expressions entered by the user. An {@link gpal.parsers.SCPNParser ANTLR-generated parser} is used to build up an {@link gpal.parsers.ast AST} (Abstract Syntax Tree) for each expression. The grammar is based on Java's grammar concerning logical and arithmetical expressions.

Type system

Understanding how it works

There a three major operations done by GPAL:

Loading and preparing the model
Checking the model
Inlining the model
Generating the C++ code

Loading and preparing the model

Starting with a new {@link gpal.SCPNet SCPNet object} the underlying XML document is analyzed recursively. During this process each XML element becomes represented by an object implementing the {@link gpal.Entity Entity interface}. Examples: {@link gpal.Place Place}, {@link gpal.TimedTransition Timed Transition}, {@link gpal.SubstitutionTransition Substitution Transition} or {@link gpal.modules.Module Module}. After loading an XML file like this, the model must be prepared before doing anything else with it. (I.e. the {@link gpal.Entity#prepare prepare() method} must be called.} During preparation phase the following actions are taken:

imported libraries are loaded
expressions are parsed
referenced tokentypes are looked up
instantiated modules are looked up

Checking the model

A prepared model can be checked. (I.e. the {@link gpal.Entity#check check method} can be called.) There are various checks that will be performed recursively on the whole model; some examples are:

existence of referenced place names
existence of referenced module names
uniqueness of names in a namespace
syntax checks on all expressions according to their respective grammars
type correctness of all expressions

Inlining the model

After checking a model it can be inlined. All parameters are replaced by their respective values. All module instances are replaced by substitution transitions containing the selected implementation as one replication. An inlined model can be stored to an XML file.