Subject: Re: SDL-News: Spontaneous transition, signal delay. Question.
From: Rick Reed TSE (rickreed#tseng.co.uk)
Date: Mon Sep 01 1997 - 11:19:51 GMT
The originator of this message is responsible for its content.
-----From Rick Reed TSE <rickreed#tseng.co.uk> to sdlnews -----
At 17:42 +0100 28/08/97, Yury Chernov wrote:
>Could somebody explain the logic, which is behind the spontaneous
>transition and signal delay concepts. According to my understanding both
>are random in their nature. If so, there should be some random generator.
>What are parameters of it? In other words, what is the probability to have
>a spontaneous transition in the state? Or what is the expected delay value?
The language definition does not define the probability or the expected
delay. This is similar to the ANY construct that does not define the
distribution probability for the result (it is just random). A spontaneous
transition is interpreted after a delay that many have "any" positive value
(or zero).
The rules for delay on channels are a little more complicated. It is
reasonable to assume that a channel delays every signal it conveys by the
same amount - this is the easiest situation to model, but may not in fact
match reality. In real systems different signals may be delayed by
different amounts due to transmission effects. However, the language
definition requires that signals never overtake one another. Thus if one
signal is delayed by (say) 10 microseconds, and the next signal is output 2
microseconds later - the delay on the second signal cannot be less than 8
microseconds (otherwise it would arrive first). The cumulative delay (and
ordering) when one delaying channel joins with another delaying channel
will also have to be considered.
>Does this implementation dependent (differ for different products - SDT,
>GEODE)?
The consequence of the current language definition is that implementations
may be different, and therefore COULD behave differently depending on
implementation, but still conform to the SDL definition.
>Is it possible to control this somehow?
Not directly in SDL. Individual tools, models or implementations will have
different ways of constraining the actual performance of systems.
The issue that needs to be considered for the language definition, is
whether such constraints could be (optional) extensions of constructs like
NONE, ANY and delay. Such constraints would limit the range of behaviours
that conform to an SDL definition, and would make the assessment of
perfomance easier. The disadvantage would be to make the language (and
underlying model) larger and more complex.
For small systems some work-arounds can be envisaged:
1) Replace delaying channels by a pair of non-delaying channels connecting
to a block containing a process that inserts the delay. The use of channel
substructure and a block type could be considered.
2) Replace spontaneous signals with timers with variable timeouts.
-- 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