Peter Watney wrote in a message to Robert Bilton:

 PW> {at}MSGID: 3:620/243.71 3110c1ed
 PW> {at}REPLY: 3:711/953 56ab7838
 PW> Monday January 29 1996 14:33, Robert Bilton wrote to Peter Watney: 

 PW> So for successful communications you have to both control the
 PW> manner in which the Proggrammable Interrupt Controller (PIC or
 PW> 8259) handles interrupts from the Serial Port vis-a-vis the timers,
 PW> the disks, etc, but you also have to control the way in which the
 PW> particular Com Port that you have connected to the serial device is
 PW> set (speed, stop bits, etc, etc) and handles the data passing
 PW> through it in both directions.

or you could do the sensible thing and write for the fossil spec. bnu and
x00 will safely handle up to 4 serial ports while digiboard (at least) have
released 8 and 16 port fossil drivers for their mutli-port cards.

 PW> There are BOOKS written on the subject!

indeed and i have a couple. after getting stuck into them and also buying
async professional, i decided that a fossil unit i got from the pascal echo
was going to do a better job!!! also if you intend to be doing any serious
programming involving comms, my STRONG recommendation is to develop your
application as a number of concurrently running finite state machines. this
is the ONLY way i have found which lets you handle the complexity of
handling multiple concurrent event-driven tasks. i have written software
which handles multiple serial ports plus the keyboard without a) losing any
serial data and b) any detectbale delay in keyboard response. Still, seeing
as the main loop was executing 9000 times/second while handling 2 serial
ports on a 386 sx-20, there should have been no delay.

ciaou 4 niaou

---