[sxa]

It has been observed that the macos/x64 build times are slower than some of the other platforms.

Since we have removed macos from the release matrix from Node 27 we should consider whether we want to continue building it in the Ci.

Considerations:

• It is now an experimental platform in 27 and so build failures should not block things
• It will give us a good heads-up as to whether new PRs work which will make backporting to release lines less risky
• Since it is close to being the most significant bottleneck as a platform it could speed up the PR test pipelines in some cases
• We may be able to reduce our Orka usage
• If anyone wishes to continue to build the runtime themselves then they can have confidence that it will work.

Opening here initially for discussion in the next build WG before opening to the wider community.

Another option would be to still build it in the daily jobs now that we have that option from the recent PR at #4518

Ref:

• macOS future strategy #4317
• build: downgrade Intel macOS support to experimental node#65427

[MikeMcC399]

This should also be coordinated with citgm testing. See nodejs/citgm#1203

[sxa]

This should also be coordinated with citgm testing. See nodejs/citgm#1203

The citgm-smoker job should use the same version selector script as this PR is modifying, so if we get macos15 added into the citgm-smoker job to be aligned with the nobuild version then I believe this should automatically apply there too when merged.

[ryanaslett]

The orka cluster has 2 x64 machines, capable of 2 vms each, we usually have 3 for testing and 1 for release.
If we reduce this to mostly just "on demand" and releases, then we can probably eliminate one of the x64 nodes in the cluster, and potentially request another arm machine to expand our testing footprint on that side.

[sxa]

@ryanaslett Thanks - I'll split that out into a separate issue to discuss that reallocation of machines in the build WG meeting and I'll add some thoughts there.