Jerome Kelleher

That's helpful, thanks. Would you mind pasting in a larger example (no need for a drawing) that shows what the current sorting of segments is, and what the required output...

Right. So, I guess to do this we'd need to update the in-memory data structure to use an avl tree sorted by left coordinate, and then squash as you go....

You're right, it's definitely closely related @gtsambos. It would be *excellent* if we can implement it using ``map/find_ancestors``! ps. Let's talk about the census event soon --- I do actually...

So, perhaps the thing to do is to add in the 'census' nodes separately then, as I can imagine this being useful without the pollarding step. Handling the mutations is...

Related to #382

Yes, cutting off the top is much easier and simpler since we don't have to worry about samples. I'll make a new function for and an issue to track. Since...

Yep, agreed. That's basically what I've done in #2240

SGTM - would need some considerable replumbing though I fear, as there's no allowance for this sort of thing in the C API.

Have you some pointers to what we can expect in terms of C API breakage @benjeffery? Nice to know what's coming up.

None of it obviously affects us from a quick glance, so hopefully we'll get through without changes :crossed_fingers: