counter example not working

[busFred]

I am trying out temporal actions example. I am using popf installed with planutils.

The following are the error I am getting. Would you please describe how you set up the planner?

/home/fred/.planutils/bin/popf  /tmp/domain--296499-cG32eWZhL1LS-.pddl /tmp/problem--296499-tyrdfYu6vs0r-.pddl 
Number of literals: 1
Constructing lookup tables:
Post filtering unreachable actions: 
Recognised a monotonic-change-induced limit on -1.000*var0(counter)
- Must be >= -3.000
Pruning (hello-world) - never appeared in initial RPG
Initial heuristic = -1.000
;; Problem unsolvable!
; Time 0.12
Planner found 1 plan(s) in 0.549secs.

planner output is just a blank.

[jan-dolejsi]

Hard to say what is offending it. Usually the combination of negative preconditions and TILs or something like that.
No, i did not use popf or optic. And the SLB planner that we use is not yet hosted for general public.
I assumed such a simple pddl would work.

[busFred]

from Slack community, someone said negative condition does not work well with optic planner. I have to reformulate hello_world to avoid negative condition. Meanwhile, commenting out over all (not (notbusy)) is necessary but doesn't seem to affect the generated plan. Not sure I fully understand the purpose of over all

(:durative-action hello
     :parameters ()
     :duration (= ?duration 3)
     :condition (and
         (at start (notbusy))
         ; (over all (not (notbusy)))
     )
     :effect (and
         (at start (not (notbusy)))
         (at end (increase (counter) 1))
         (at end (notbusy))
     )
 )

[jan-dolejsi]

Negative pre-conditions do not work well in combination of other constructs - like TILs - in popf. It prints a warning.

Sure, the example can be reformulated that way.

The role of the over all condition is not necessary in the simplistic case. As long as all actions that modify the predicate also check for its specific value in their pre-condition consistently, the over all is not necessary. Let's simplify it to a form that would work with at least one planner. A pull request will be welcome.