Jerome Kelleher

This sounds like a topic much more suited to the "tskit tutorials" site. See proposal here: tskit-dev/tutorials#10 Shall we migrate this issue to the "tutorials" repo?

OK, will do once we have buy-in for a few more people on the tutorials repo side.

Ruff is quite opinionated about getting people to use f strings etc isn't it? What were the errors? LGTM though, happy to switch. Given the noisiness we may want to...

Super neat! That won't be very efficient as it's calling back to python, but we could easily implement the node spans in C using this approach

> Confusing! I suppose it might possibly be due to the C code calling back out to the python summary function?!? Otherwise I don't see why that code should take...

I don't think there's a right answer here tbh. The VCF spec says it wants 1-based coords, but I'm not sure anything bad will happen if there is a zero...

> I think the issue is only that tskit allows a nucleotide at position 0, which is a much smaller deal. I agree with Peter here. Msprime **generates** sites that...

(As a follow up, we could easily add an option to msprime to disallow sites with position=0, which I'd be happy to discuss over there)

Yes, @mufernando and @bhaller are correct in that tskit's interpretation of genome coordinates is zero-based (it allows a zero coordinate position). However, we do actively store coordinates in there that...

I think it's ok to make this breaking change as it's best seen as a bug fix. I do hate to break people's code, but this seems like the lesser...