Leo Torres

When there is no canonical way of doing things, we should offer both ways OR implement whatever is easiest for the time being. > What should we do if there...

> That would be super nice to be able to do without reindexing all the edges in H3. I agree, but the question is, after executing this code, what do...

If we ever implement `__add__` then we must make sure that this is commutative so that `H_1 + H_2` always gives the same as `H_2 + H_1`. If we implement...

I was just thinking about this. The NodeStat functionality is going to be merged soon, and something I don't have implemented yet (but should) is precisely the capability to compute...

Oh good point. Well I think in that case we can have two methods: `H.nodes.degree.ashist()`, returning bin centers and absolute/relative counts, and then `H.nodes.degree.asdist()`, returning bin edges and normalized counts....

Coming back to this, I think I would like to implement the following, with the default output type being a dataframe with one row per bin: ```python3 H.nodes.degree.asdist() # bin_center...

> * I'm still unsure about the call syntax that is long with so many dots (not just for this one probably). Any thoughts? I agree I don't like it...

**EDIT**: I no longer advocate for option 3, see below. ~~Just to be a bit more concrete, here is an example skeleton of `EmptyHypergraph`:~~ ```python3 class EmptyHypergraph(Hypergraph): def add_node(self, node):...

Thank you for your answer. > * right now, `H.nodes.degree` and all other similar stats functions would give some uninformative error like `None has no method degree()` or something, right?...

Can you please comment on a couple of those ways? So we can start a discussion comparing and contrasting them.