i intend to land this on Saturday unless anyone objects before then.

as discussed on IRC, if you want to document the current situation, i agree with @markj that the section on update-packages needs more detail, specifically about how it only adds changed packages to the new repository.

can we just have a single target (for example, make packages) which either creates or updates the repository? if necessary there could also be a make clean-packages target but i'm not sure this is needed, you can just delete the repository if you want that.

actually i see the file was previously installed in the ftp package (not ftpd) which seems also dubious as that's for the FTP client, which won't be using PAM at all. so i think the change is still right but i'll clarify the commit message before landing.

i've tested pkg upgrade with this change and it is not confused by replacing the hard link with a separate file as far as i can tell (e.g., it doesn't replace /etc/pam.d/ftpd with the contents of /etc/pam.d/ftp).

per comments, remove the link

if we're intending for this file to be used by ports, FreeBSD-ftpd may not be the right place to put it in the first place; perhaps it should live in -runtime with (most of) the rest of the PAM configs?

In D50339#1148758, @kevans wrote:

Looking at the target from a broader view, it seems like it only exists because we don't have a notion of per-group LINKS.

In D50339#1148701, @des wrote:

I would prefer fixing the mk scripts, if that's not too much work?

use FTPTAG_ARGS

add an UPDATING entry

also initialise the variable to false

i'll update the commit id when landing this

review fixes

although i don't think rc.subr should depend on bsdconfig in general, the real purpose of this change is to let us move bsdconfig out of the FreeBSD-utilities package in pkgbase.

review fixes

closing this in favour of D50019. thank you everyone for your input on a surprisingly complicated issue :-)

In D50292#1147383, @zlei wrote:

RIP is quite a simple routing protocol to be implemented, and can be found from lots of routers. It is still available from modern routers, even modern networks prefer other routing protocols.

this seems fine to me, but add a license and perhaps don't MFC it until the DESCRIPTION has been improved a bit, for example, "The .Nm driver supports Marvel 88E61xx switch chips". no one really knows what these do anyway since the documentation is a trade secret.

fix an error in the route6d.8 manpage

i appreciate the concern but i don't know what else we'd use "fd" for - it's not like we need a package of file descriptors (in case people run out). since all the tools start are called fd* and floppy disks were historically /dev/fd* this seems like a good name.

will bump Dd before landing

appears correct based on UPDATING and also my memory of how you're meant to set this.

if this lands i will send a heads up to current@ and pkgbase@ since we don't really have a good way to handle adding new packages right now.

tweak a comment to be more informative

fix rcmds ucl; this was converted wrong due to its different Makefile syntax.

new version:

review fixes

update __FreeBSD_version. this got lost during the rebase.

new version of link_addr().

mention std::ignore in cdefs.9 for C++

• add doc in cdefs.9
• remove result_use_or_ignore_check and replace uses of it with nodiscard

okay, in that case POSIX gives me no other option than to approve this change