David Widmann

Yeah, I think DistributionsAD is actually a good example for why ideally this should be part of PDMats but not be defined in a separate package. Hopefully DistributionsAD will be...

I think I can review (and probably merge) a PR in PDMats if you want to prepare one @willtebbutt.

> Like NNlib here, where it's guarding a mutating path... seems likely to be the right thing even for non-Zygote reverse AD. I guess this specific use case could also...

I just ran into the `sqrt` issue. Do you think you'll be able to finish and merge this PR soonish, @sethaxen? Or would you like some help here?

>> zero(::NotImplemented) throws a NotImplementedException > > This feels like it should be a ZeroTangent() As it is one of the things that is in fact safe to do even...

Can you open an issue in ReverseDiff? (In general, I'm not surprised, I think I came across some issues with the macros in ReverseDiff a while ago but didn't have...

Maybe the documentation issue is an issue of ReverseDiff (as well)? I.e., maybe the issue should be moved to ReverseDiff?

A docstring exists but is not included in the documentation: https://github.com/JuliaDiff/ReverseDiff.jl/pull/196

Didn't want to chime in here initially, but I still think it's more accessible if one provides instructions for how to format the code with Julia (and possibly the most...

I wonder if it would be sufficient to use https://docs.github.com/en/actions/learn-github-actions/events-that-trigger-workflows#pull_request_target instead of `pull_request` or if this would break the reviewdog action. If it doesn't work it would be useful if...