Seems to work fine now! Thanks! :)

Since Malte agreed, I merged this into the public default branch.

The patch does not affect the translator output on the standard benchmark tasks
and works for the given test instance but it is in a part of the code that we do
not touch very often. If you still encounter problems on other test instances,
feel free to reopen the issue.

Is there a fix I can test? I found few other similar problems (some even without
either) where I can test it...

I don't really know the old code well enough, so if you're happy with the change
and it passes all the tests, I'm happy to see it merged without further review.

The fix is ready for reviewing. It does not have significant impact on the
translation time and the output on the existing benchmarks tasks is unchanged
(except for a few trucks-strips instances where the differences are presumably
due to the timeout of the invariant synthesis triggering at different states).

As Gabi wrote, the "either" isn't the main issue here, but I should point out
that it is not supported at all. We're curently missing the error reporting to
clearly indicate all cases where the PDDL input is not supported, unfortunately.
If you want to track the progress on that, it's issue323.

The attached example is wonderfully simple. Thanks a lot!

The invariant finder has some issue with constants. If I include the elevator as
an additional action parameter, everything works.

This looks like a problem in the invariant synthesis. The translator finds a
mutex group
[Atom at(?X, n8)]
which implies that there can only be one passenger at floor n8.
I will have a look into this.

Apart from this, Fast Downward does not fully support (either ...) types (cf.
http://www.fast-downward.org/PddlSupport). But this appears not to be the source
of the problem.

Hi,
for the attached problem, no solution has been found even though it has a solution.

With any of these small changes of the problem, the solution is found:
 - delete one of goals
 - delete unnecessary fact (at p3 n8) from init
 - change second goal (at p6 n8) -> (at p6 n5) 

It seems that there is something wrong in translator, since the generated SAS is
empty.

Thanks for help!