Jerome Kelleher

This does look great @nspope and it's a really important piece of functionality. I wonder if it'd be worth talking through at one of our community meetings? These happen every...

I don't want to derail the conversation here, so no need to respond, but I just wanted to put the question into people's heads: should this type of functionality be...

This sounds great @nspope - looking forward to hearing more about this one Wednesday, and have a good discussion about how we get the infrastructure into tskit. I am sympathetic...

You mention counting tree topologies @nspope, so pointing out the [combinartorics stuff](https://tskit.dev/tskit/docs/stable/topological-analysis.html#sec-combinatorics) in case you're not aware of it.

What would you call this? ``ts.max_time`` has a pretty clear interpretation. I guess ``ts.max_tree_node_time``? What are the applications?

I think we should keep ``ts.max_time`` simple - it is the maximum of the times of the entities in the data model (and conversely for min_time) - ``np.max([np.max(ts.tables.nodes.time), np.max(ts.tables.mutations.time)])`` If...

> I assume that the time of the virtual root, since it is not actually stored in a table (I think?) is not considered to be part of the data...

Adding an ``order`` argument similar to the Tree.nodes() method would solve the majority of this (``ts.nodes(order="timedesc")``) I think. The ``order`` defaults to increasing node ID. Getting the k nodes that...

Changed issue title. ```python def nodes(self, *, order=None): order = "id" if order is None else order if order not in ["id", "timedesc", "timeasc"]: raise ValueError(...) # do stuff ```

This is so simple that maybe what we want is a just method to return the ``dosage`` instead? So, we'd do something like ```python site_sq_corr = [] for variant1, variant2...