Updated the domprob parser.

[haz]

Closes #25

[ssardina]

Checking this, finally! I spotted two side glitches, fixed them in c46d65e here directly, too much hassle to open issues for that; hope that's OK with you @haz

Now I am running the tests; det succeeds all, but normalization seems to be failing; checking that now:

[haz]

Welcome back! Totally fine to commit directly on. 25a915e fix the issues?

[ssardina]

OK @haz , I am almost done. I need to fix the test cases bc the new pddl system seems to output things different than before. For example, it adds object type to all non-typed constants:

This made me think our comparison between two PDDLs whether they are equal could be too strong and fragile. As soon as there is a char difference, a test will fail. So any re-formatting of the PDDL would break comparison between two otherwise same PDDL files.

I am thinking whether I should change the unit test cases to not care about the exact formatting of a PDDL, whether it has a space here or two or 10, similar with \n

Will fix and improve this tomorrow and get this out of the way. Thanks for the patient Chris! 👍🏽

[haz]

Ya, I'd say skip spacing for comparisons would be best. Let the pddl package be finicky with that

[ssardina]

The normalizer is dropping the :non-deterministic requirement, but it should not. The normalized PDDL would still use oneof:

I am puzzled why @haz decided not to include that requirement, there is explicit code (see screenshot) to drop it. Maybe it was wrongly imported from a previous determinizer?

[ssardina]

OK good, now all tests are passing:

• Sync the expected files to what the new pddl lib gives (e.g., adds - object)
• made equivalence between 2 pddls independent of syntax in expected file by parsing and printing via pddl lib

[ssardina]

OK checked it all and looks good now. Fixed and improved a few things, including aligning all tests for determinizer and normalizer.

All good to go Chris @haz !

[haz]

Ya, that requirements removal was an oversight, and almost certainly a copy/paste from the determinizer [here]. Nice catch, and thanks for the review!

[ssardina]

@haz , should we bump the version now that is aligned with the new pddl?

I was going to do it to v0.1.8 , but I thought I would check with you. I think we may want to do v0.2.0 as they are changes based on a new pddl and will not be "backwards" compatible

[haz]

Ah, good point. Let's go with 0.2.0, as it'll be a bit broken with the update.

[ssardina]

Done! upgraded to 0.2.0