Planner input and output requirements

[Nathan Robinson]

Hi,

I was wondering about the I/O requirements of the planners.

Are there special requirements for the command line arguments to be
taken by our planners, or can we specify the arguments required?

What are the naming requirements for the file where the resulting
plans will be stored?

What action should be taken (in terms of writing to the output file)
in the case that a planner times our runs out of memory?

Cheers,
Nathan Robinson

[Charles Gretton]

Please figure this out for yourself...

[Malte Helmert]

On May 26, 3:18 am, Nathan Robinson <nathan.m.robin...@gmail.com>
wrote:

> Hi,
>
> I was wondering about the I/O requirements of the planners.

We will put up detailed instructions later today, but for now, here's
the answers to your questions:

> Are there special requirements for the command line arguments to be
> taken by our planners, or can we specify the arguments required?

The planner should be callable as
plan domain.pddl problem.pddl output.soln
where domain.pddl and problem.pddl are the inputs and output.soln is
the name of the file to which the result plan should be written. If
your planner has a different calling syntax, it's probably easiest to
wrap it in a simple shell script.

> What are the naming requirements for the file where the resulting
> plans will be stored?

See above -- the output filename will be passed to the planner.
Special considerations apply for anytime planners -- these should
create a sequence of outputs like output.soln.1, output.soln.2, etc.

> What action should be taken (in terms of writing to the output file)
> in the case that a planner times our runs out of memory?

Nothing needs to be done. The planner will be killed by an external
mechanism, and we will record that it didn't solve the problem if we
don't find an output file.

Best regards,

Malte