Planners are ready, some tracks are not

[Malte Helmert]

Dear IPC-2008 group,

we spent most of last week getting the submitted planners to compile
and run, and I am very pleased to say that all planners are working
now. In many cases, we had to make smaller changes to the source code
to get the planners to compile. In particular, no fewer than 10
planners used flex/bison parsers derived from FF's source code which
don't work with current versions of these tools. Fortunately, we could
fix this. There were also a number of issues with 64-bit compilation
-- so many, in fact, that we've decided to compile all the planners in
32-bit mode by default, even though we might run one or two
competition tracks on 64-bit machines. If you are certain that your
planner is 64-bit ready and would prefer to have it run in 64-bit mode
on appropriate machines, please notify us. Please note that we will
set a memory limit of 2 GB per planner run, so there is no need for a
64-bit address space.

As was to be expected considering the enormous effort required to
implement a planner, there were a number of participants that dropped
out of the race. Comparing the total number of registrations we got
with the current numbes, we get the following statistics:

* sequential satisficing track: 18 planners registered, of which 9
still remain
* sequential optimal track: 18 planners registered, of which 9 still
remain
* temporal satisficing track: 11 planners registered, of which 5
still remain
* temporal optimal track: 5 planners registered, of which 1 still
remains
* net benefit satisficing track: 8 planners registered, of which 1
still remains
* net benefit optimal track: 6 planners registered, of which 3 still
remain

Of course, a single planner is not enough to run a track, so
unfortunately we have to cancel the temporal optimal track and the net
benefit satisficing track. In both of these tracks, there were a large
number of last-minute withdrawals, and until today we hoped that we
might be able to run the net benefit satisficing track at least, but
it wasn't to be. We are sorry for the two teams that invested a lot of
effort in making their planners ready for these tracks.

What happens next?

First, there are a number of copyright issues to be resolved with some
of the planners. I'll discuss this point in a separate email.

Second, we will run some preliminary experiments to see that the
planners produce valid plans, and that optimal planners indeed produce
optimal plans. We want to automate these experiments as much as
possible. To avoid the overhead of emailing each team about the
outcome of testing runs, we will use the IPC wiki for communication.
To keep this communication private to the organizers and the
respective teams, we will use access-protected wiki pages. This means
that all teams (or at least one person in each team) need to create
accounts on the IPC wiki if they haven't done so yet.

So, if you haven't done so yet, please create an account on
http://ipc.informatik.uni-freiburg.de/ and notify us once you have
done so. Please also tell us if you have already created a wiki
account with a user name from which it is not clear that it belongs to
you. You don't need to contact us if you already have a wiki account
and it is easy to tell that it yours from your user name.

Best,

Malte