Alistair Miles

Confirmed, only a manual call to ``s3.invalidate_cache()`` without any arguments clears the cache: ```python In [1]: import s3fs In [3]: s3 = s3fs.S3FileSystem(client_kwargs=dict(region_name='eu-west-2')) In [4]: store = s3fs.S3Map(root='zarr-demo/store', s3=s3, check=False)...

I've also been looking into similar behaviour on GCP. It seems like the `user-scheduler` is intended to improve the packing of user pods into nodes, as described [here](https://z2jh.jupyter.org/en/stable/administrator/optimization.html#optimizations). In other...

In case anyone else stumbles on this also working on GCP, I also found that some kube-system pods were blocking scale down of user and/or worker nodes, particularly kube-dns and...

Thanks @sanjaynagi, just to support that having an implementation of G123 would be very useful. I imagine that the implementation would be very similar to H12, except that we need...

Also just to note that whatever approach is taken it should be able to handle multiallelic variants, i.e., variants where the allele indices are not just 0 or 1.

This is an issue for us as well, trying to build a jupyter book with ipyleaflet maps in it and hitting this. A release would be much appreciated 🙏🏻 cc...

Btw we are instantiating a `Map` with an initial zoom level and we still see this, so perhaps #1068 will not fix this after all?

Just wondering if anyone has raised this with the colab devs? Looks like they need to update the way that widget state is stored in the notebook JSON.

Hi Martijn, Thanks for the reply, I've been using that pattern to add INFO fields to the header and INFO values to variants and it's working well. The problem I...

Hi James, apologies, I've had my head in another problem. Thanks for scuba quick response, I'll try out your patch ASAP and get back to you re the API. On...