Gabriel Baraldi
Gabriel Baraldi
Do we know which one broke it?
What is segfaulting here? Because I don't see how that breaks stuff. We just register functions with the JIT
But if we remove the assertion (it seems slightly weird anyway)
The `skipmissing` regression is real if anyone wants a deeper look. I really wish BaseBenchmarks had an option to ask for the LLVM IR of the function being called. 😉
I think this is ready to merge. @vtjnash and @vchuravy any objections?
Yeah, lets do it.
Yeah this seems thread unsafe from a quick look.
These benchmarks are incredibly sensitive to random things and this is not release blocking so I'm taking it the milestone but assigning myself
Cthlhu disagrees with code typed and ```julia Base.infer_effects(*, (String, String)) (+c,+e,+n,+t,!s,!m,+u,+o) ```
Since we are making things threadsafe I would look at all `@async` uses because we want to deprecate it.