by jasper » 30 April 2017, 11:53
1) there was some trouble in the fact that controllers are eliminated by the COCO 3.0 solver job scheduler, which in itself is ok (as the controller itself gets evaluated not by its calculate, but in the total function evaluation), but left the dependency on a virtual input stream for set point unaccounted for in job order determination as well as its Jacobian contribution. Fixed (a pretty large solver overhaul to fix that), thank you for reporting. The case also exposed a sequencing issue for parametric jobs, which is now solved. Fixes are available via CUP.
Not sure of whether this was introduced by your artificial example, but in your example case for the second controller, the measured value at solution must always be equal to the controlled value. Hence, you can, in this case, eliminate the controller, directly control the feed rate from the other controller and eliminate the measure unit.
2) not clear on what you mean. What exactly does not update? Can you send me a screen shot?