Thanks for all of your hard work on this package and considering the option. The architecture reexporting Plots is not a problem unto itself (I don’t think… we used to have many weird precompilation issues with StatPlots, Distributions, and Plots… but tough to know if any of that was due to the reexport or general precompilation madness).
If users can’t go using Plots higher up in a notebook and then using StatsPlots later on when they realize they want to plot a distrbituion, it is not the end of the world by any means, I just need to be clear in the lecture notes about it. But hopefully the issue is now fixed!
Because StatsPlots brings in dependencies on the Stats packages, and people on light-weight systems may not want that.
So the separation originates back from when DataFrames was a monster that pulled in half of the Julia package ecosystem (I’m exaggerating for effect of course). So the DataFrame functionality was split out to keep Plots light (Plots is very light in terms of deps).
Now with Tables that distinction is not important anymore, and that’s why we’ve rethought the purpose of StatsPlots to be associated with the StatsKit packages.
That’s pretty neat, had no idea - feels like this might be something to add to the readme? While the examples show that to use the package you should just do using StatsPlots without using Plots, the first sentence states " intended to be used with Plots.jl:" which might increase the likelihood of people opting for using Plots, StatsPlots (as I have done until now, without running into errors fwiw)