Frederik Rietdijk

I was just debugging this as well today, and it occurs with Dask/Distributed 2.6.x. With older versions there is no issue.

On the mailing list I proposed a meta attribute for this "Proposal new meta attribute: meta.permitAutoUpgrade". > For the Python package set I would like to introduce a new meta...

@asymmetric typicallly it is perfectly fine to update patch level changes, but with minor and major that's not always the case. Default is indeed `true` or `"major"`. In case of...

Also, with Python it becomes increasingly important now that many packages no longer release Python 2 compatible versions. Then we need to restrict certain expressions.

Yep, that's it. Though we might also want it only for X>=1 with the version X.Y.Z, [because](https://semver.org/#spec-item-4) > Major version zero (0.y.z) is for initial development. Anything may change at...

I wonder then whether we need a new primop for applying a patch (a content-addressable path) to a content-addressable path producing a new content-addressable path. That could be used with...

It would mean there is a Nix builtin you call to apply one or more patches. How this is implemented, I don't know. Maybe it invokes `patch` impurely just as...

I think it should be created, but be opened as a draft with preferably a note. I always keep them open as a draft, since otherwise a new PR is...

Similar issue with `hvplot.xarray`. Note there it fails eventually with a `NotImplementedError` in `pandas`. It's a bit unfortunate it also fails even when not using that index. Related issue on...

I think this could be done before opening a PR. E.g., cypari was merged, and then reverted. By checking whether the changeset is already in the tree we know that...