[Info-vax] Real time, was: Re: Volatile, was: Re: yet another sys$qiow question

[Simon Clubley]

On 2015-08-28, JF Mezei <jfmezei.spamnot at vaxination.ca> wrote:
> The reason I had been given for not using ASTs and using a loop for hard
> real time was that they could ensure timing on a known event loop while
> AST delivery was unpredictable since something else could stop AST
> deliveries and delay processing etc.
>
> (Just recounting what I was told).
>

Even if you use a polling loop you still have to account for your code
being stalled by an interrupt at some random point in time and you need
to allow for the time it takes for that interrupt to be serviced.

> Similarly, on space station, they used a very old military slow network
> type that is polled at predictable times because they felt ethernet was
> unpredictable because you might have a collision. (not considering the
> fact that with Ethernet 100 times faster than that military protocol
> would have no problem handling the load generated by those devices).
>

It doesn't matter how fast Ethernet is; the root problem is that it
doesn't have deterministic behaviour.

It's like the people who say something is "real time" because it goes
really fast; that isn't what real time is about.

Real time is about deterministic behaviour, not speed, and the same is
true both in software (in a RTOS for example) and in hardware (with the
networking example above being one relevant example.)

Simon.

-- 
Simon Clubley, clubley at remove_me.eisner.decus.org-Earth.UFP
Microsoft: Bringing you 1980s technology to a 21st century world