Jerome Kelleher

> Hm, well we don't necessarily remove the site, since what if there's a polymorphic mutation at it still... We'll only remove it if ``filter_sites`` is True and the number...

Yes, this should be doable I think, and I agree it's a good option to have.

Is this still something we want @petrelharp?

Huh, I thought we had an option for this in simplify. It's probably quite tricky to get right in the general case, I'm certain I thought about this at the...

I think returning ``inf`` is semantically fine and is inline with the time of the virtual root being +inf. We've also defined the branch length of a root as zero,...

It'd be more troublesome to start messing around with missing data I think, as we'd have to start defining different behaviour on sample and non-sample nodes. Just having a time...

Very cool! > is this a good name? a good python API? I also thought about just making weights an argument to genetic_relatedness, which would call one method or the...

> I'm not sure how to track that down - can we run valgrind on the python tests? No, Python gets in the way. We'd have to reproduce the error...

Thanks @petrelharp, this is a weird one. Reliably fails on the 10th seed, but not if we just run that seed on its own. I'll poke around, but it nearly...

I think it's worthwhile putting in some tests to cover the remaining code paths now - this will have to be done later anyway, and it might unearth the problem...