I wrote a resource manager for an A/D for QNX4 many years ago. Here was the
I client process would “register” with the resource manager. The
regestration process would indicate channel numbers, minimum and maximum
thresholds and a proxy number. If a threshold was reached a proxy (would be
a pulse in RTP) was sent to the client. Then the client would poll the
server to find out the exact new values for any channels necessary.
The threshold concept was import to reduce IPC traffic. If a channel was
indicating .42 volts currently you may say don’t wake me unless it gets to
…40 or .44. That way you don’t have to be woken up due to insignificant
fluctuations in input voltage.
In my design, when a threshold was reached the client has to send a new
registration message. Otherwise the client might get flooded with messages
once a threshold is solidly reached.
Bill Caroselli – 1(530) 510-7292
“Jun” <firstname.lastname@example.org> wrote in message news:email@example.com…
I am writing a Analog-to-Digital converter (10 channels @ 16bit) resource
The client must poll (read()) the resource manager all the time in order
get the data from it.
How can the resource manager “inform” the client
as soon as the data are ready without waiting for its polling? Because
critical data must be processed immediately
by the client.