Soliciing suggestions for application design

Byron Morgan lazypointer at yahoo.com
Mon Mar 24 23:17:54 CET 2003


Hope I don't have to load Linux for this, but it is not out of the question.

Cost of an occasional improper correlation is low, but of course I want
fairly high reliability in the output. I can do some validation of results.
The remote device is not to be messed with, as it is an ancient PDP 11-84
running a Motorola data-controlled radio system. One could not get away with
modifying the protocol.

I have loaded Dev-C++ and am looking into doing the serial comms with that,
but so far have not discovered any serial library for it. Any tips in this
area?

Byron


"Oren Tirosh" <oren-py-l at hishome.net> wrote in message
news:mailman.1048500878.31495.python-list at python.org...
> On Sun, Mar 23, 2003 at 05:35:30PM +0000, Byron Morgan wrote:
> > Oren, thanks for your input.
> >
> > The processing of captured data can be done without critical time
> > constraints, but it is important to get the timestamps right, as this is
the
> > only way to reliably correlate responses to polls.
>
> But what is the cost of an miscorrelation? If the answer is $50 I suggest
> you should get a true real-time operating system like QNX or Linux with
> real-time extensions. You might see zero failures on a non real-time
system
> in many hours of testing system but you can never know for sure.
>
> If the remote device is also under your control I suggest modifying the
> protocol to add some kind of ID in the poll that is echoed in the
response.
>
>     Oren
>






More information about the Python-list mailing list