ssardina-research / app-trapper Goto Github PK

I can make the offline version, but online (see I have renamed main-traps.c to main-online.c in #1):

[ssardina@Thinkpad-X1 src]$ make online
gcc main-online.c -o pp-online -D__TOOLDIR__="\"`pwd | sed 's/src/tools/g'`\""
main-online.c: In function ‘create_domain_file’:
main-online.c:480:20: warning: too many arguments for format [-Wformat-extra-args]
  480 |       fprintf(out, ")\n", j); // the j is not used at all, no place in formatted string
      |                    ^~~~~
main-online.c: In function ‘run_planner’:
main-online.c:597:7: internal compiler error: in subspan, at input.h:68
  597 |       snprintf(tmp, MAX_STR_LEN, COMMAND_LPG_1SOL, seed);
      |       ^~~~~~~~
Please submit a full bug report,
with preprocessed source if appropriate.
See <file:///usr/share/doc/gcc-9/README.Bugs> for instructions.
make: *** [makefile:7: online] Error 1

Wehn I try the online solver:

[ssardina@Thinkpad-X1 example]$ ../src/pp-online obj.pddl init.pddl predicates.pddl acts.pddl graph.txt 123 lpg
Checking plan: emptyplan.tmp
Plan executed successfully - checking goal
Goal not satisfied
Plan invalid


Failed plans:
 emptyplan.tmp 
Running trapper as follows: /home/ssardina/git/soft/planning/app/app-trapper.git/tools/lpg -f pfile.pddl -o domain.pddl -out soln.tmp -n 1 -seed 1692901013 -cputime 60 -nobestfirst -inst_with_contraddicting_objects -same_object

 Unknown option: -same_object entered


LPG-td-1.0 SETTINGS:

This is because the commands in def.h have the -same_object when calling LPG. It is strange because they are in the original LPG indeed, but that does not seem to be recognized by the lpg binary!

What is happening here? If I delete that option from the commands, then it works!

This is not a problem, it is my misunderstanding of C and I am curious. ;-)

When I compile, say offline solver, I get many errors of this sort:

I want to understand what is happening here. My C is not great for sure. :-)

It is true the arguments to the function are char *:

So if I understand well, that means that fgoals is a pointer to (one) char. However, when it is called:

Now that future_goals is defined in the Node struct as:

char future_goals[MAX_CONN][MAX_STR_LEN]

I understand that an array of something in C, is just the address to the first element of the array.

But future_goals is not just an array of charof size 500 (MAX_STR_LEN): it is an array of an array of chars...

But the compiler seems to expect a `char (*)[500]’

Why it is not telling me that it should be char (*)[100][500] (since MAX_CONN = 100).

@lchrpa , I suspect this is trivial for you...

ops, I should leave it open so that I don't forget to have this and the other (non-working unlucky example)

Originally posted by @ssardina in #7 (comment)

@lchrpa , can we remove the j here? (main-online.c)

There are to main C files in src/: main-trap.c and main.c.

Which is the right one ?

@lchrpa , we have this function:

and we call it like this:

I have replaced it with:

Agree? Why did it work well before?

Would be nice to clean-up the repo and code a bit so that it can be distributed to others that want to use it and test experiments.

Basically we want a README that can be followed step-by-step, from setup to compile to run.

System does not seem to run with lpg option:

[ssardina@Thinkpad-X1 example]$ ../src/pp-traps obj.pddl init.pddl predicates.pddl acts.pddl graph.txt 123 lpg
Checking plan: emptyplan.tmp
Plan executed successfully - checking goal
Plan valid
Final value: 0

Successful plans:
Value: 0
 emptyplan.tmp 


FAIL

Hi @lchrpa

With the examples you provided, I have done quite a lot of clean-up, both at source level and documentation.

I have left log run files *.log in each example.

One thing that I am still trying to remember is the role of LPG when we use use DFS+ or LMCUT. So, the paper says:

So it seems that even when we use offline with DFS+, the LPG planner is called and this is true as it prints things like this:

; Command line: /home/ssardina/git/soft/planning/app/app-trapper.git/tools/lpg -f pfile.pddl -o domain.pddl -n 1 -input_plan soln.tmp -seed 125 -same_objects  

However, I can see those LPG calls also in the online solvers when using DFS with and without traps.

And all this happens on both of our examples that do NOT have loops in the APP, they are both linear.

Do you know what is happening or what am I missing @lchrpa ?

; Command line: /home/ssardina/git/soft/planning/app/app-trapper.git/tools/lpg -f pfile.pddl -o domain.pddl -out soln

 Command line: /home/ssardina/git/soft/planning/app/app-trapper.git/tools/lpg -f pfile.pddl -o domain.pddl -n 1 -inp
ut_plan soln.tmp -seed 1234 -same_objects