Stephane Richard <[log in to unmask]> writes:

> Hello everyone,
>
> First I would like to thank you for your prompt and very insightful
> replies.
>
> As a following to the last email, amongst the replies I have received, an
> issue (that is mandatorily important to me) has been risen.
>
> Now it seems that Ada's Real Time package needs not be question upon it's
> stability properties.  However, I may need to question it's performances or
> it's native design.
>
> What I need is a timer, that can regularly perform jobs and processes at a
> very regular and highly small interval of 1/10000 if a second.  The issue
> that was raised was one that it wasn't sure Ada could perform such a small
> time interval.

This is purely an implementation issue. Look in your compiler
documentation for Annex D; the accuracy of a delay statement using
Real_Time.Clock must be documented.

Or, to put it another way:

This is not a language issue, it is a hardware and operating system
(or compiler run-time system) issue. If you have a hardware clock that
is good to 10e-5 seconds, and you have a run-time system that has
latencies less than that, then you have a system that meets your
requirements.

> What I would like to know is if anyone has programming (recently)
> for real time and what the current performances and specifications
> are for the Real_Time package?

For any language:

On Unix, 1 second.

On Windows NT, 10 milliseconds.

On VxWorks PowerPC, 100 microseconds (I think).

On custom, fast, real-time hardware and OS; as fast as you need it!

--
-- Stephe