Markus Dreseler

@julianmenzler - this is what you saw before, too, correct?

We should probably add error handling to the existing pexpect test.

@Bensk1 @mweisgut Did I miss something?

Yes. Still wanted to bring that up, as that is quite important.

See AbstractOperator::deep_copy(). Updated the top post.

Hmpf. Bloody strings. Do you have numbers on how much we can save for those if we ignore the stability of the sort result?

wolfsort could be another stable alternative: https://github.com/scandum/wolfsort

Ah, right. It's late and I just finished the last wine bottle from the tasting :/ I wonder if we can come up with a sorting mechanism that doesn't rely...

What if instead of iteratively sorting the data column by column, we first independently sort all columns. Instead of returning a sorted `PosList`, we write the position into a `vector

> (ii) Complexity: the stable sort is pretty much on par with the non-stable sort, but we add another sort phase. I don't know if that will be advantageous for...