Specification of Reactive Hardware/Software Systems - Electronic ...

9.3 Formal Syntax 263
In almost any complex reactive hardware/software system messages or conditions can
be identified that interfere with the normal course of behaviour. In general, these
messages or conditions have to be dealt with immediately, no matter what other activities
are currently being carried out. To specify such complex behaviour, POOSL provides
an abort statement and an interrupt statement. The difference between an abort and an
interrupt is that the former aborts the current course of behaviour whereas the latter only
temporarily interrupts the current course of behaviour. The abort operator is similar to
the disabling operator of LOTOS [vEVD89].
Processes contain internal data in the form of data objects (also called travelling objects)
which are stored in instance variables. Data objects are private to the owning process, i.e.,
process objects have no shared variables or shared data. A process can interact with its
data objects by sending messages to them. When a process sends a message to one of
its data objects, its activities are suspended until the result of the message arrives. Data
objects themselves cannot send messages (except for replies) to a process object.
When two processes communicate, a message and a (possibly empty) set of parameters
is passed from one process to another. The parameters refer to objects which are private
to the sending process. Because processes do not have any data in common, it does not
suffice just to pass a set of references to the data objects, as in traditional object-oriented
languages. Instead, the objects themselves have to be passed (whence the term travelling
object). This means that a new set of objects has to be created within the environment of
the receiving process. These objects are deepCopies of the data objects involved in the
rendezvous. The concept of travelling object was first introduced in [Ver92].
Next to processes, POOSL supports clusters. A cluster is hierarchically built from processes
and other clusters and acts as an abstraction of these. The constituents of a cluster
are composed by parallel composition, channel hiding and channels renaming. These
combinators are based upon similar combinators originally used in CCS [Mil80, Mil89].
Processes and clusters are grouped in classes, just as data objects. Members of classes
are called instances. A process is an instance of a process class, and a cluster is an instance
of a cluster class. Each class can be parameterised. Parameters refer to data objects and
are used to initialise an instance of a class.
9.3 Formal Syntax
This section describes the formal abstract syntax of POOSL. It is based on the language
of data objects of the previous chapter. We assume that the following sets of syntactic
elements are given:
CName p process class names C p ¥¡ ¡ ¡
CName c cluster class names C c ¥¡ ¡ ¡
Chan communication channels ch¥¡ ¡ ¡
EPar expression parameters P¥¡ ¡ ¡