Lukas Hergt

Thanks for pointing this out! This might resolve (part of) #22.

If cloning and pip installing still works, then no objection from my side.

Regarding the TEX labels, these are not so much metadata of the entire dataframe, but, more precisely, each is connected to one corresponding column. As such I really like the...

Currently this PR is not set up to merge into our master branch, was that intentional? As a consequence CI is not working as supposed to, if it was not...

@williamjameshandley and @Ormorod, I've encountered these failing Windows tests multiple times now. It's not a deterministic bug, i.e. sometimes simply rerunning the same test will work... Also difficult to debug,...

Hi Stefan, for completeness it's probably good to mention that there of course is the option of calculating the standard deviation with `sample.x0.std()`. That is probably the best thing to...

Hi Stefan, the `fastCL` function is a really neat approach avoiding a KDE and the choice of a bandwidth. Seems to work well for large quantiles (e.g. 0.95), for smaller...

> **Edit 2:** @lukashergt at what sort of rate were you seeing failures? the version above this comment has failed conda windows 3.10 2/2 times, below has succeeded 10/10. To...

> @lukashergt I notice that the conda windows install includes pyqt, which is an optional matplotlib backend. Could this be affecting it? Possibly, although from #244 it seems like the...

> @lukashergt I think you've added a test for this in #243 ? I've added a test with a `cmap` kwarg to `test_samples.py`, yes. Note that I added this to...