Re: SDL-News: Object-orientation in SDL


Subject: Re: SDL-News: Object-orientation in SDL
From: Birger Møller-Pedersen (Birger.Moller-Pedersen#ericsson.no)
Date: Thu Dec 03 1998 - 07:53:29 GMT


Glenn Lewis wrote:

> The originator of this message is responsible for its content.
> -----From Glenn Lewis <Glenn.Lewis#utas.edu.au> to sdlnews -----
>
> Hi!
>
> I am interested in object-orientation in formal methods. I wonder if
> anyone
> can tell me of papers/reports etc assesing the object-oriented
> features of
> SDL?
>
> I am rather new to SDL (I have only done some introductory reading),
> but
> what I understand of how OO is supported in SDL is below. I would
> appreciate any answers/corrections/comments etc and or confirmation
> that I
> have got it right (...I am particularly unsure of the final sentence
> ...)
>
> Instead of classes SDL speaks of types. The types can be instantiated
> and
> refernces are used, but there is a one-one relation between references
> and
> instances (i.e. you don't have the general references between objects
> as in
> an OO language).

You have Process Identifiers that correspond to object references - they
are, however, not typed.

> SDL supports a form of single inheritance, where it is
> possible to replace or redefine components but this is limited, for
> example
> you can only replace a transition that is declared to be virtual (does
> this
> mean that once such a transition is refined, it cannot be further
> refined?

SDL has the notion of redefined (not refined) and finalized. A redefined
is still virtual, so that it can be redefined, while a finalized can not
be redefined

> is it only possible to redefine virtual components of processes?).

No, of any kind of type that has some kind of "content", that is block
type, service type, procedures, but not signals.

> There is
> no way to guarantee that the behaviour of the subtype is consistent
> with
> the behaviour of the supertype, i.e. there is no concept of
> behavioural
> inheritance or subtyping of some other OO languages (has this ever
> been
> addressed or is it just left up to the specifier?).

This has been discussed, but to my knowledge there is no language that
guarantees you
consistent behavioural inheritance. The behaviour inheritance of SDL is,
as most other oo languages, on a syntactic level.

/birger




This archive was generated by hypermail 2a23 : Sun Jun 16 2013 - 10:41:40 GMT