Carlo Cabrera

MacPorts does as well. I think they're able to build LLVM for up to 10.10, but that relies on MacPorts-specific patches (they ship a special `libSystem` wrapper). We can backport...

Had a peek inside the runner, since `swift` only has a handful of dependents so it shouldn't be taking this long. It's been stuck testing `vapor` for about 5 hours.

Linux dependents need revision bumps, I think. But that may be better done separately.

Static linking isn't great, but if upstream is going to make it the default, then we may as well go with it.

For context: the idea was for `brew reinstall` and/or `brew upgrade` to avoid undoing a previous user-requested `brew unlink`. Discussion of that here: https://github.com/Homebrew/discussions/discussions/3555#discussioncomment-3439381

> I think we need to instead fix the reasons people want individual non-keg-only formulae to stay unlinked rather than allow it. One common reason is conflicts. One example we...

> This feels like something solved by either: > > - make things keg-only so they don't conflict and `brew link` is sticky > > - remove binaries/symlinks so they...

> Note: the "`brew link` is sticky" bit is already true. I thought it would be; that's good to know. > > Alternatively, we could also try out removing the...

`watchman` fails to build on Monterey with ``` Undefined symbols for architecture x86_64: "folly::f14::detail::F14LinkCheck::check()", referenced from: folly::f14::detail::F14Table::rehashImpl(unsigned long, unsigned long, unsigned long, unsigned long, unsigned long)::'lambda'()::operator()() const in libeden_service_thrift.a(eden_types.cpp.o) ```...

Failed on Catalina as well. Superenv doesn't seem to be adding `-march` flags, so that might be why. Logs: [watchman.tar.gz](https://github.com/Homebrew/homebrew-core/files/9640272/watchman.tar.gz)