Lilith Orion Hafner
Lilith Orion Hafner
Thanks! This mostly LGTM but I would like to keep a test for 1.0 on Mac.
Something to consider before merging is that this does increase CI latency in the event that ubuntu/1 + windows/min is slower than Docs
> why does the Base version work? Base uses `ignore_warnings=true`
Here's a slightly more minimal MWE on 0.11.3: ```julia julia> ChisqTest([0,100], [0,1]) Pearson's Chi-square Test ------------------------- Population details: parameter of interest: Multinomial Probabilities value under h_0: [0.0, 1.0] point estimate:...
The API seems reasonable. Performance of writing `NaN` and `Inf` under `allow_inf` is a bout 2x slower with this PR than before (probably due to the changed writing implementation). IDK...
Adding `quoted_inf_mapping` and `underscore_inf_mapping` feels like standard proliferation, other than that this seems reasonable. It's plausible the custom mapping case could be faster but I don't think that's blocking as...
Definitely in the docs; and at that point it doesn't really matter much which.
CI failures look to be the same as the failures on main.
I don't have a piece of code to paste. I'm also not that invested in the API here; Happy to hear @quinnj's thoughts.
Yes, IIRC it was for performance reasons.