Phillip Alday

"Technically non breaking" is not great for something near the root of the dependency tree. Fixing the ambiguities here is straightforward (and certainly no more difficult than "just" replacing a...

Removing undocumented unexported "clearly" internal type aliases was viewed as breaking: https://github.com/JuliaStats/StatsBase.jl/pull/840

I still think this is technically breaking, though I've come around to the idea that PValue should not be a numeric type. We should also document this a little, even...

@devmotion I think that was very much the intent of @ararslan -- proposing a longer-term solution that should not be implemented in this PR but which could inform our strategy...

I also wasn't sure that this was the best path -- I really did it just to see what would happen. And I also agree that the (re)compilation overhead in...

Needs a codecov secret added to the repository -- @ssfrr can you do that?

@ssfrr Still getting no token found 😦 : https://github.com/ssfrr/DeepDiffs.jl/actions/runs/8556959474/job/23491069276#step:8:31

I think a nice middle ground here might be to support parallel computation the same way `plyr` does -- take more-or-less seamless advantage of any parallel backend available, but require...

Thinking about this more: I think `parallel=TRUE` should use pass that argument untouched onto `llply` while `parallel='auto'` will attempt to set up an appropriate backend using distributed memory via `doParallel`,...

In my fork, there is some parallel support, but until it's integrated with upstream, the API is subject to change. There's also support for testing multiple coefficients / terms per...