Sebastian Berg

Maybe the same as the local failure, adding `-- -s` to the spin run locally gives me: ``` collecting ... ERROR: Flag 'coordination_agent_recoverable' was defined more than once but with...

Well, had a conda setup, but bumping jax from 0.7.0 to 0.7.1 indeed fixes the issue for me locally. Should just try rerunning the pixi lock to update that. (and...

Ah, I am having some problems with the `spatch` when updating (`failed to extract metadata for package 'spatch'` dunno). Without your fix for spatch, adding `!=0.7.0` indeed resolves me to...

> From the docs, I'd summarize spatch as essentially a combo of multipledispatch (or is it single-dispatch?, unclear from the docs - what happens with mixed array types?) Right. I...

> then the one the user adds needs to be tried first. Everyone has very different opinions about this... Right now, builtin ones have _slightly_ higher priority (to avoid accidentally...

I'll make a `spatch` issue to allow ignoring `requires_opt_in=False` unless the name is in something like an `allow_always_active=` list. (Even if SciPy doesn't care, I suspect others will.) That would...

> like this would not allow us to dispatch based on array shapes, correct? It's possible and implemented, just not first class, because it doesn't make sense to try and...

I need to test this a bit better with cupy still, but took some time yesterday to change the pattern to what I would intend it to actually be longer...

I'll consider this ready for review but there are still 1.5 issues: * ~Locally, this should switch from using `cupyx` to using `scipy` for `freqz`. But that makes the tests...

I think the helpful first step may to be convert the file to C++ to continue in steps from there. These custom macros seem a step in the wrong direction,...