I added the test for duplicate operator names to issue215 ("translator should
loudly complain about invalid PDDL") and close this issue here.

I don't remember, but hg log shows no commit in that time that changes anything
in VAL, so I guess I didn't update it.

I think this was a "bug" because the translator did not generate a warning then
(I don't know if it currently does). The thing about VAL and INVAL was just to
figure out what other tools do in this case. Since they classify this as an
invalid input, I thought we should also classify this as invalid input. But if
Fast Downward does the right thing even if there are duplicate names, we could
also close the issue.

Florian, did you update the VAL version in the Fast Downward repository after
the bug fix?

(And why has this issue priority "bug"? Do we aim to fully validate the
correctness of the input tasks?)

And another test.

(Testing issue tracker after OS upgrade.)

I sent a bug report via email in July and Derek responded with a fixed version
of VAL (4.2.10), so I assume the code on github already includes this fix.

Hi Florian, now that VAL has a public bug tracker at its github site, can you
report the bug there?

Please send a bug report to the VAL maintainers, too.

The translator should generate an error if the PDDL file contains two actions
with the same name. The semantic of such cases is not clearly defined by PDDL
but VAL and INVAL both cannot verify generated plans for such domains: VAL seems
to consider only the first instance of the action, INVAL detects the duplicate
and reports the domain as invalid.