Subject: SDL-News: Spontaneous transition,delay - "just random"
From: Rick Reed TSE (rickreed#tseng.co.uk)
Date: Tue Sep 02 1997 - 17:22:20 GMT
The originator of this message is responsible for its content.
-----From Rick Reed TSE <rickreed#tseng.co.uk> to sdlnews -----
Yury Chernov has asked me to expand on the clause "it is just random" that
I used in my previous answer.
Because SDL is intended for specification, there is sometimes a need to
express the idea that a value from a particular set of values is chosen.
The SDL expression:
can be either True or False. If this expression is evaluated many times
during the interpretation of the SDL, there is no (language) requirement of
fairness between the two values. An implementation that always interprets
the expression as True, conforms to the rules of Z.100. Similarly where
NONE has been used, an implementation that always interprets the
spontaneous transition conforms to Z.100 rules, as does one that never
interprets the spontaneous transition, or one that interprets the
transition after a varying delays. The delays in delaying channels can vary
in the same way (except that signals must never overtake one another).
It was perhaps misleading of me to use the word "random", because this
usually implies something about the distribution of values. Z.100 states
"From a value returned by an Anyvalue-expression no assumption can be
derived on other values returned by Anyvalue-expression." The language
determines the range of values, but not their distribution.
Consider some SDL in which the only "non-deterministic" feature is a single
ANY(Boolean) expression. If the objective is to investigate all the
properties of the SDL, then a whole class of behaviours needs to be
investigated covering the cases where the expression is always true, always
false, alternately true and false, 50% probability of being true, 1%
probability of being true, true one thousand times and then always false,
and so on. It is clear that even this simple example covers an infinite
number of cases. Normally something more is known about the distribution of
values, and this would be reflected in a simulation. SDL does not currently
provide any mechanism for specifying this.
Although this is the meaning of ANY according to Z.100, most practitioners
(including myself) usually assume that the AnyExpression values are
Yury Chernov, also replied
>In my opinion, SDL is a simulation language (to some extend) and will gain
>if there is a possibility to define different distributions for delays,
>timers, even tasks.
>I realize that it could be built by the user by means of operators or
>procedures. I'm thinking about special library as in some other languages.
>The second thing to add is the sensitivity analysis. Also one the
>simulation features. However this must be much more difficult to implement.
>What do you think about expanding of SDL to the simulation direction? It
>would be very nice to know your opinion.
This was discussed at the last SG10 meeting under the topic of Time and
Performance Modelling, and agreement was reached that this is an important
issue and needs to be further investigated, and therefore is now on the
list of topics for study under Q.6/10. Further contributions have been
invited. The issue was raised by a contribution from
Carlos.Rodriguez#imag.fr and Joseph.Sifakis#imag.fr via CNET. If you are a
member of the SDL Forum Society, you can obtain the paper as TD33 of the
April/May SG10 meeting from the Society server.
-- 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 email: rickreed#tseng.co.uk http://www.tseng.co.uk ftp://ftp.tseng.co.uk/tseng/
-----End text from Rick Reed TSE <rickreed#tseng.co.uk> 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