Debugging macos CI failures

[Jasha10]

Based on PR #2909.

Currently looking at homebrew-related MacOS CI failure https://app.circleci.com/pipelines/github/facebookresearch/hydra/15662/workflows/d3e1e4ea-301a-4b33-b899-61f2867925ea/jobs/162011

[Jasha10]

There's currently a "brownout" on circleCI's macOS Intel Support. According to that link, circleCI Intel resources will be sunset on June 28, 2024. This is causing the current failures of all macOS jobs.

[omry]

Oh, this is starting to make sense.
So the cash was cashing intel binaries and they were fetched on M1 hardware?

[omry]

About Nevergrad:
I am actually aware of an issue with nevergrad and the latest numpy.
They fixed it in facebookresearch/nevergrad#1613.
Until they release a new version compatible with the latest numpy I think we can just pin numpy.

[Jasha10]

So the cash was cashing intel binaries and they were fetched on M1 hardware?

Yes 😄

[Jasha10]

About Nevergrad:
...
Until they release a new version compatible with the latest numpy I think we can just pin numpy.

Ok, sounds good!

[crassirostris]

@omry @Jasha10 do you need help? I'm happy to help to make Hydra work with the newer versions of antlr

[Jasha10]

@crassirostris yes please! If you can finish the PR to get the CI green, that would be great.
But I didn't think that antlr was the problem...?