Jerome Kelleher

It's not a trivial thing to do (with the GSL dependency) but would happily accept a PR if someone is motivated.

I'm a bit unclear as to what the node type should represent to be honest, but then my understanding of the structured coalescent is pretty hazy. Let's see if others...

@benjeffery any idea what's up with CircleCI here?

Can we remove the bintrees dependency or are we still using it somewhere? I think the main issues here are with the verification code - how much have you run...

Dammit, didn't see that coming :frowning_face: I'm surprised that there's so many adjacent segments with identical nodes but different ancestral_to values, I thought that would be rare...

Can you push the changes to the C code please @GertjanBisschop - I'd like to have a look at the segment squashing logic

Hmm, maybe we're not defragging often enough. Can you set it up so that we always defrag the segment chain after msp_merge_two_segments (probably just set ``defrag_required=true``) and look at those...

This is a real head scratcher... The extra segments all seem to be late in the simulation, when they should be pretty widely separated within the lineage (I would have...

Yes, that looks about right for your hypothesis. I'm still surprised there are so many adjacent, squashable segments with different ``ancestral_to`` values - how can this happen?

Right, I guess that's it. At times I've wondered if actually keeping track of these values is worth it at all. In principle, you could just look through all segments...