On Fri, May 09, 2025 at 02:22:54PM -0700, Dan Mahoney (Ports) wrote: > > > > On May 9, 2025, at 07:13, Mathieu Arnold <[email protected]> wrote: > > > > On Thu, May 08, 2025 at 04:29:26PM -0700, Dan Mahoney (Ports) wrote: > >> All, > >> > >> I’m doing some work on Jared Mauch’s efforts to make Mailman2 work under > >> Python3 (because mailman3 is….a problem child that I don’t want to talk > >> to)…. > >> > >> And I notice the original port uses something in > >> /usr/ports/Mk/users/fakeroot. > >> > >> Documentation on what this does is a little thin. Could someone tell me > >> what cases this is required in, and what it does? > > > > All port building is done as an unprivileged user, so fakeroot, > > basically, is a wrapper that helps run command that expect to be ran as > > root. > > > > For example, if the install process does, say, `chown root /some/file`, > > as it is running as nobody, it will fail. Now, if the install process is > > running through fakeroot, that chown will basically be a noop, but it > > will not fail, and the install process finishes properly. > > > Aah, now I see. So, this is of benefit to the port building on the build > clusters that’s used to eventually build .pkg files — not for people who (I > think as mentioned on -questions) are doing their port building by doing the > classic cd /usr/ports/category/foo; make install. > > That was the nuance i had missed.
Yes, but it is the other way round, ports have to be able to package as an unprivileged user, it's for the remaining people who run the unsupported `make install` on a live system as root of all things that it's not needed. -- Mathieu Arnold
signature.asc
Description: PGP signature
