Jerome Kelleher
Jerome Kelleher
Just to update, I've submitted for a talk at the RSE conference. I'll report back here with any further news.
> The one tricky part of this implementation was making sure that seeking to an arbitrary position in the tree sequence works properly. We only want to add branches that...
I'm happy to merge this @lkirk, do you want to do some final tidy-up first?
That's interesting - I didn't realise people were using edge metadata. Out of curiosity, can you show us the edge table to see what's in there @TymekPieszko? ```python tables =...
I see, that's interesting - Relate is doing something useful with the metadata. Good for them! Dropping the metadata is most easily done like this I think: ```python tables =...
Excellent! I think you just need to cast to (int) for the printfs, give the commits a bit of a squash and it's ready to go in.
Some quick answers here @awohns - We should aim for consistency with the stats API, even if this means breaking the current interface, I think. - Supplying both ``windows`` and...
Everything to do with (span) windows should be done *exactly* as it's done for the stats API @hyanwong - there's mechanisms in there for all sorts of stuff.
I'm having trouble understanding what the output is - can we see some simple example code here?
Just the simplest, most naive version of time-aware GNN in Python would be really helpful now.