Cecil Curry

> Part of a startup doing something completely different these days! That's wonderful! Bay Area startups are so fun. Aside from the death march, perpetual crunch, and tenuous paychecks, I...

Maybe we should just go back to the hideous threadlocal approach. It's trivial for @beartype to generate simplistic statements that hook into `jaxtyping` during type-checking like: ```python import jaxtyping def...

**Hearken!** I couldn't help but cogitate about this issue all day while bicycling. That's when you know an issue is spicy. Notably, the constraint that `state` be replaced on only...

**Hah, hah!** Surprisingly, @beartype's behaviour *is* supposed to happen here. We're doing something right for once. The reason why will both disappoint and shock you. Technically, PEP 484-compliant stringified forward...

**...guh!** You got me. I can no longer deny the horrifying truth unfolding before my very eyes. This is incontrovertible evidence in a court of QA law: > Moreover, the...

**...lol.** I confess to pure ignorance. Inscrutable non-determinism between JAX's NumPy layer (i.e., `jax.numpy`) and NumPy itself are disconcerting, baffling, and triggering. I feel deep-set fear in my soul. Luckily,...

As an intellectual curiosity (and mostly a note to myself), I couldn't help but uselessly cogitate: > "*Can* @beartype even circumvent this JAX issue? Might be impossible. That would suck."...

[Quax](https://github.com/patrick-kidger/quax) looks so cool! Congrats, @patrick-kidger. You've made yet another awesome runtime QA thing. Jeez. What gruesome problems won't that guy solve next? 🤔 And now allow me say... ##...

@patrick-kidger: Fellow QA wizard, I summon thee! Firstly, apologies for not properly finishing up feature request #544 ("Official `jaxtyping` integration") with you. I have excuses. They are bad excuses –...