Tue Feb 23 17:54:18 EST 2016
How to get better timing constraints?
Erlangs model of proceses and messages works very well.
On a (bare-bones) microcontroller, state machines can take the role of
true proceses. Often it is possible to put constraints on the depth of
a call stack. However, dealing with message queues in
memory-constrained systems is harder.
What happens when the queue is full? Easy: a task should block.
So how to mix this with strict timing constraints?