Patrick Kidger

Ah, that's a really good point! I hadn't appreciated that. I'll keep that in mind when I work on simplifying the definition.

I think mypy doesn't support this yet: https://github.com/python/mypy/issues/17547 You could try using pyright instead, which does.

Depends on the solver! One could put anything behind that interface. As a practical matter I think all of the built-in ones for Diffrax should be safe for this (but...

Yup, this is a known issue. I don't have a nice way to fix this right now -- this is quite a complicated corner of jaxtyping! -- but I'd be...

So I think the underlying `__cause__` should be visible in your traceback, just by scrolling up a little in your terminal? The per-argument typechecking should probably have the correct underlying...

So for this one I'm actually planning on doing my `Module` refactor fairly shortly. (As my private testing with it has so far gone smoothly.) For that reason I'm inclined...

These are an interesting set of ideas! I can totally see why these are attractive for power users. Points (1) and (2) I think probably impose too great a learning...

Right, so what you're seeing here is that in the first snippet, the autogenerated `__init__` method will set `self.x = None` as an instance attribute. Meanwhile in your second snippet,...

Let me know if you figure it out! Other than that -- this has a lot of duplicate code with `error_if`. Probably this is most easily done with a Boolean...

Yeah, I think that's the right place to do it! (Although FWIW I think some stuff like stacklevel may be incompatible with JIT? Or maybe you need to just preemptively...