Validate that GDL rules aren't malformed

[samschreiber]

There have been reports about malformed rules in game rulesheets like the following:

  (<= terminal (not (sheep_exist)))  ;; instead (<= terminal (not sheep_exist))
  (<= (base (step 1)))               ;; instead (base (step 1))

These should fail validation in the GDL validator. Currently I don't think we have anything that checks for this during validation; we should add something to do that.

[AlexLandau]

Side note: the GdlCleaner transformation should fix both of these (along with a few other common problems), but this is certainly something we could add to validation. It would be easy enough to check.

[samschreiber]

Awesome, thanks! What I'd ultimately like is to have any game that passes validation be good enough to publish (permanently!) on a public game repository with no manual review. We're not there yet, but that's my ultimate goal.

[AlexLandau]

I have this implemented, but a lot of the test case games violate this, and it's not clear whether they're meant to test cases of games written like this or not (e.g. in test case 1a). Which is to say, do I "fix" those game descriptions, or update the StaticValidatorTest to say they're supposed to fail?

[samschreiber]

Is the malformed GDL essential to what's being tested? If so, indicate that those test cases shouldn't pass the StaticValidator since they're deliberately malformed. Otherwise, if the malformed rules could be rewritten in a way that they validate properly, but still test what they need to test, they should be rewritten.

I definitely think we should have test cases that include malformed rules, to make sure our players can handle them; but that doesn't mean that all (or most) of the test cases have to include malformed GDL, since many test cases will be testing things unrelated to malformed GDL.

[AlexLandau]

See #30