Jan Kwakkel

The behavior change should not matter. In this code, a new set of rows is added to an existing data frame, so there should not be any NA columns anyway....

Thanks for this! I will take a deeper dive at this, hopefully, tonight or otherwise over the weekend. I quickly checked the code this morning and have a few additional...

I have made some choices in light of all the comments and feedback. Feel free to discuss if you disagree. In short: * added NetworkGrid * Renamed various classes *...

> compared to my code from last week, this is still a bit slower. I'll try to investigate later. I fixed the performance issues. The code base is mainly complete....

> I don't think this isolates it too much, this is something that has been on several wish lists. I just meant that it keeps all experimental code within a...

> If you agree conceptually I think we shouldn't already break this concept at such an early stage It happened largely because of me trying to chase down the performance...

I have started adding tests for this code. Coverage is not complete yet. I have focused on the initialization of OrthogonalGrid, HexGrid, and NetworkGrid, as well as the neighborhood calculation...

It's 400 lines, so 25% of the current space module. But indeed, it is getting to a point where it is a bit unwieldy. So I am okay with a...

I am running into 3.9 vs 3.10 now as well. I can fix the type annotations in #1994 to make it work with 3.9 or first move to 3.10, so...

I agree with @Corvince. Let's give UV a few weeks to sort out the initial problems before we consider shifting. Also, what is the problem with a CI that takes...