04:42 <jaeger> Personally I have no objection to removing libbsd/libmd if possible with cmake/meson or explicit disables... I think the only port I use that actually hard requires it is opendkim

12:23 <farkuhar> The benefit of a non-standard install location (as libbsd was configured in 3.7 contrib) is that most ports will not need any extra configure/cmake/meson flags, they'll avoid linking to libbsd because they don't see it in the standard search paths. Having to insert new flags, for every port that might auto-detect libbsd, entails more work for the maintainers.

15:51 <jaeger> Yeah, that was exactly the reason for the original non-standard path

17:31 <farkuhar> As darfo pointed out on Saturday, "It would and it did" solve the problem of eager linking, to install libbsd in a non-standard path. That argument in favor of the /usr/opt location is equally valid whether libbsd is in 3.8 core, or 3.7 contrib. In either case, the work for port maintainers is lessened when they don't have to actively opt-out of linking libbsd.

17:35 <SiFuh> json-c ignores system headers and poaches libbsd and fails under MUSL never tested it under glibc.

17:53 <farkuhar> While we're on the subject of non-standard paths, opt/iputils after the update to 20240905 started installing four new man-pages, but under /usr/man rather than /usr/share/man. I noticed this odd location while looking for possible updates to the prtverify whitelist.

20:42 <jaeger> hrmm, what happened to the 'groups' binary?

20:42 <jaeger> (in 3.8)

21:44 <darfo> jaeger: https://github.com/shadow-maint/shadow/issues/1010 says it moved to coreutils/busybox in >= 4.16

21:44 <jaeger> ahhh, ok. Thanks for the tip, hadn't gone looking yet

21:47 <jaeger> simple enough to add groups back to core/coreutils, I'll take care of that later when I have a chance

21:49 <darfo> yep, --enable-no-install-program :)

21:49 <darfo> just found it

21:55 <jaeger> yeah