Echelon Neuron C - Manual (Page 65)

Neuron C Programmer’s Guide 53
heater_failed = FALSE;
// heater device received update
}
The nv_update_completes Event
The nv_update_completes event evaluates to TRUE whenever an output network
variable update or poll either succeeds or fails. An example of testing for
network variable update completion is shown below:
#include
#define C_TO_K 2740
IO_7 input ontime invert clock(2) io_temperature_sensor;
network output SNVT_temp nvoCurrentTemp;
when (nv_update_completes(nvoCurrentTemp))
{ // latest temperature has been sent out
ontime_t
sensor_value;
// send another update
sensor_value = io_in(io_temperature_sensor);
nvoCurrentTemp = (sensor_value * 221) / 642
+ 211 + C_TO_K;
// tenths of a degree,C
}
If a program checks for nv_update_completes or nv_update_succeeds for any
network variable, the program is said to use comprehensive completion event
testing. See Comprehensive Completion Event Testing on page 55 for the rules
you should follow.
Synchronous Network Variables
When an output network variable is updated, the Neuron firmware ensures that
the most recent value assigned to an output is propagated and received as an
event by any connected input network variables. Thus, if multiple updates are
made to an output network variable within the same critical section, only the last
value assigned is ensured to be propagated and received as an event at the input
network variables.
A critical section is defined by the boundaries of the executing when task. After
the when task completes, the most recent value is scheduled for propagation.
You can specify that
all updates to an output network variable (within the critical
section) must be propagated and received as events by using the synchronous
subclass of network variables.
However, for a synchronous output network variable, each updated value is
immediately scheduled for propagation, which requires an application output
buffer for each scheduled update. If all output buffers are already in use when an
update to a synchronous output network variable is scheduled, the Neuron
firmware enters preemption mode to attempt to complete pending transactions
and thus free in-use application output buffers.
As a result, updating synchronous output network variables can cause your when
task to execute in a non-deterministic fashion. See
Preemption Mode on page 55
for more information about buffers and preemption mode.