Thank you

I would prefer KBI compat under a COMPAT_FREEBSD14 option closer to D56660. The userland changes require updating to the latest 14.x to have a working jail, which is a reasonable thing to officially ask of users (not stable). Yet a "vnet jail" should "just work" even with a newer kernel shouldn't it? I would think there is an implied compat layer for jail features. I have a 9.0 jail that works fine without vnet because jails are quite simple otherwise. So the question is, what is the promised compat support for vnet jails? Is it only kernel==userland? That's very strict considering jails give such leeway to upgrade slowly otherwise. I think we should consider codifying this support somewhere if it's not already.
Also this isn't limited to jails. We're asking people to upgrade to latest 14.x before 15.x, or to disable ipfw on upgrade, all of which requires reading which users commonly don't do on upgrades.

+1 for this. D54763 wasn't sufficient for my jails use case as getosreldate() isn't reliable.
We should maintain compatibility for this kind of thing _at least_ for 15.x branch and delete it in 16.

D44809 removed mmap too.

-C for builds makes a lot of sense.
I think we should remove the limit all together. It was added in originally because mmap was being used in commit de264c2bd89b931303db40fb8a50c9ee5ed8f858.

Even if it's a bug in node, the symlink option installs a symlink to /usr/local/openssl/cert.pem, and now the package does not do that. It's a simple hack to keep locally but there is a behavior change here that was probably not intended.

Why? I assume certctl is expected to do the right thing but it doesn't. The port was installing a symlink to /usr/local/openssl/cert.pem but now nothing lands there and node breaks.

<removed draft comment I meant to remove>

I ended up moving the LIBWRAP stuff to the tcpwrappers patch but it's still there.

Merging into D53273

Merging into D53273

Merging into D53273

Merging into D53273

D53268 covers this. The servconf changes belongs there. It will come in with the ports change.

I have this staged for commit in ports for D53268. The version.h removal I'll bring in eventually. It's part of a lot of glue so can't come out easily.

I have this staged to commit to ports. Thank you.

Can we get a test added for this?

In D55191#1294886, @des wrote:

I'm still waiting for someone to explain what this is supposed to mean:

Since the test assumes a 250 ms response time, the test delay can be 1 second instead of 3.

Thank you! I ran it through poudriere's test suite with no problems.

Nice work

• Fix handling of case 3

This still has a problem with the 3rd case.

# 3. Common sources in SRCS.
#         SRCS= common.c
#         SRCS.prog1= prog1.c
#         SRCS.prog2= prog2.c

The common.o.meta is not getting generated.

• Add additional comment clarifying this code is used in the parent _and_ when recursing.

It was pointed out to me that "current current" should be interpreted as "current CURRENT" which makes sense. So not a typo.

Commit message typo "current current"