Christian Muise

@marcofavorito @francescofuggitti Can I get your take on this one? I've banged my head against the proverbial Lark wall for quite some time now, and am convinced that our choice...

Ah, I see. Just leave it up to the post-grammar parsing. Allowing any number of arguments to a minus is kind of risky...put in a test to force it to...

Ya, I think sticking to the available bnf makes sense. This means handling minus in two ways, and having 3+ operands only work for addition and multiplication. Do we want...

Let's leave this open for a full fix. I'm partway through. One issue that just arose is the inheritance. Functions inherit from Atomic, which inherits from Formula. This means you...

Fair enough -- this issue, along with some of the others that have come up, stem from my first medium scale encoding task with the library. It occurs to me...

Unless there's major objections, I think we'll move forward with a companion library that focuses on advanced techniques for modelling. It will build heavily on `nnf`, and the student (👋...

Would this work as a third option? - Every variable obviously has a unique hash (as does its negation). - Every node has a hash built recursively on the (sorted)...

Ah, well I was assuming you always hash the string of an object. And variables wrapping an integer should then be fine (collisions notwithstanding): i.e., `hash(Var(-1)) == hash(str(-1)) != hash(str(-2))...

Ah, ya, I see your point. Given a generic `__str__` implementation on whatever's forced into `Var` objects, then we run into issues at the Variable level. But wouldn't what you're...

Btw, how does this entire discussion relate to the deduplication you've done? https://python-nnf.readthedocs.io/en/stable/_modules/nnf.html#NNF.deduplicate