Subject: Re: SDL-News: Delays in Channels (repeat)
bmc#informatik.uni-essen.de
Date: Mon Jan 27 1997 - 06:02:56 GMT
The originator of this message is responsible for its content.
-----From bmc#informatik.uni-essen.de (Bruno Mueller-Clostermann) to
sdlnews -----
> > I am not very comfortable with the non-deterministic delays in channels
> > as defined by SDL88. Does this mean the ordering of events at the input
> > queue of a process is purely random? After all, different sequencing of
> > events can lead to different behaviour by the state machine. So, I would
> > be forced to conclude that SDL cannot be used as a language for modeling
> > the behaviour of a system. Am I right?
> >
> > Santhosh Kumar Plakket <santhosh#pts.mot.com>
> >
> Though the delay in channels is nondeterministic, signals cannot
> overtake each other, which means that the ordering is preserved. The
> same applies when the signals are in the input queue. Signals may
> overtake each other if they use different parallel delaying
> channels between the same source and destination. Also, with
> SDL-92 and beyond, non-delaying channels may be used as well.
>
> Amardeo Sarma < sarma#gwup.org>
> ----------------------
I think Amardeo's reply has clarified the channel delay semantic. Here I
add some
comments on SDL extensions concerning delay and timing aspects. In our QSDL
(=
Queueing SDL) we have a construct "output ... delay ..." that delivers
signals
to the input queue of the destination process with a delay that may be
specified
by the designer. The delay parameter may be an arbitrary (random) function
including the deterministic (constant) case as well as the
non-deterministic case
with arbitrary overtaking of signals (if you like to do that). The latter
case
leads of course to an explosion of the state space (due to the growing
numbers of
interleavings) hampering exhaustive validation.
Explicitely specifiable delays allow for the investigation of a specified
system
under timing constraints ("timed validation"). Moreover QSDL provides
resources
("machines"), features for load modelling, and more. The QUEST tool
provides
performance evaluation and timed validation of (extended) SDL systems.
(Ref: Proc
SDL-Forum '95, Proc. of FORTE/PSVT'96)
Regards, Bruno Mueller-Clostermann
----------------------------------------------------------------------------
--| Bruno Mueller-Clostermann | Tel: +(49) 201/183-3915 (-2352) |
| FB 6/ Informatik | Fax: +(49) 201/183-2419 |
| Universitaet GH Essen | email: bmc#informatik.uni-essen.de |
| Schuetzenbahn 70, R. 504 (Hochhaus) | http://www.informatik.uni-essen.de/ |
| D-45117 Essen | |
---------------------------------------------------------------------------- --
-----End text from bmc#informatik.uni-essen.de (Bruno Mueller-Clostermann) to sdlnews -----
For help, email "majordomo#sdl-forum.org" with the body of your email as:
help
or (iff this does not answer your question) email: owner-sdlnews#sdl-forum.org
This archive was generated by hypermail 2a23 : Sun Jun 16 2013 - 10:41:39 GMT