Re: SDL-News: transition atomicity in SDL

Subject: Re: SDL-News: transition atomicity in SDL
Date: Tue Apr 01 1997 - 12:34:30 GMT

The originator of this message is responsible for its content.

-----From Rick Reed TSE <> to sdlnews -----

> (Stefan Leue) wrote:


>The notion of atomicity has been well-known...

> ... When using an interleaving semantics, which is

>a suitable and probably the most convenient way to represent

>concurrency in distributed systems, atomic events are those that cannot

>be interleaved by other concurrent events.

But SDL processes are not interleaved (according to Z.100), even (from the

quick check I have just done) in the formal semantic model of SDL (Z.100

Annex F.3). They are meant to be truly concurrent: each process exists


In any case, the formal semantics are only a MODEL of the SDL language - it

is not expected that SDL is necessarily implemented in this way - only that

the model should exhibit the same observable behaviour as the SDL.

I did know what "atomic" usually means, but then it is still open to

argument whether a transition can be divided or not. Once started the

transition will finish, but on the other hand the transition takes some

(probably small) time, so that there is a possibility that between two

outputs an event may occur in another process. Does this count as division

of the transition or not? Is it in fact meaningful to compare that sequence

of events in one process with a sequence in another process, unless and

until there is an event that concerns the both processes (such as the input

of a signal by one previously output by the other)?

It is still unclear to me whether SDL transitions can be classed as

"atomic" or not. It seems that any transition that is independent of an

external value is atomic, whereas transitions that include a non-atomic

procedure, Now or view are non-atomic.


Rick Reed, TSE Limited

13 Weston House, 18-22 Church Street

Lutterworth Leicestershire LE17 4AW United Kingdom

Tel +44 14 55 55 96 55; Fax +44 14 55 55 96 58


-----End text from Rick Reed TSE <> to sdlnews -----

For help, email "" with the body of your email as:


or (iff this does not answer your question) email:

This archive was generated by hypermail 2a23 : Sun Jun 16 2013 - 10:41:39 GMT