Probably should upstream this

Also add a gone_in_dev() in attach?

I think this is a vestige of the times when we built a separate sysroot.

• Fix missing braces

• Fix error returns

In D40519#922646, @emaste wrote:

This driver should probably be on the list of potential drivers to cull in 15.

Probably - are you willing to add the man page deprecation notice and gone_in?

This driver should probably be on the list of potential drivers to cull in 15. Other than quite a few bug fixes to ioctl handling it hasn't seen any non-sweeping commits since 2016.

Enabling this as an experiment seems plausible. It might be nice to start with excluding contrib out of the gate to avoid generating too much noise.

This feels like flavors might make sense instead of the shared pseudo ports, but seems fine as is. I might give the pseudo ports a nonstandard directory name to highlight that they are aren't really ports (e.g., devel/_wasi-libcxx or devel/.wasi-libcxx).

Superseded by D39954. Commandeering to close

I think I'd be inclined to remove it and restore it once there's a consumer. It's not hard to restore. If it's going to stay we need descriptions for src.conf.5 which are hard to write when we don't quite know what it will do.

Absolutely not.

• Use a single pkg invocation

Replaced by D39680 D39681 D39682. Concurrency limits make this not work well.

Alternatively, NETLINK could be documented and a dependency relationship created between it and https://reviews.freebsd.org/D39678.

This should be pushed to GitHub and tested as a pull request before commit. That will ensure bootstrapping various bits works across the environments we test.