Hi Simon, I'm dropping debian-mentors@d.o from the recipients list as I think it the discussion is not relevant for that list anymore.
On 08/04/2018 16:20, Simon McVittie wrote: > On Sun, 08 Apr 2018 at 08:26:13 -0600, Daniele Nicolodi wrote: >> the package is dbus-broker, a replacement for dbus-deamon. You may have >> heard of it: there has been a short exchange about its packaging for >> Debian with its developers with the Debian dbus maintainers in Cc. > > Sorry, I didn't see that conversation until now. Please use the role address > <package name>@packages.debian.org if you want to reach package maintainers: Will do, thanks. I thought I added all maintainers individual addresses at some point in the discussion, but probably I didn't... > If dbus-broker is uploaded to Debian as an optional dbus-daemon > replacement, it will definitely need to be coordinated with the dbus > source package. Having the two packages coexist is probably not going to > be straightforward to set up, and if any diversions, alternatives etc. > are going on, all maintainers of the dbus package will need to be aware > of them. I have the package ready, and it works fine without any diversion, alternatives or other trickery. It has been surprisingly easy, indeed. You can get the package here: https://salsa.debian.org/dnn-guest/dbus-broker It works fine minus replacing dbus-deamon for the user bus, which prompted the email at the origin of this thread. It needs to be done manually (see README.Debian) or installing a /etc/systemd/user/dbus.service symlink. I don't know what's the best strategy, yet. I'm happy to work with the dbus maintainers to fix any issue. Incidentally, I will be looking for a sponsor to upload the package to unstable as soon as I figure out how to solve the above minor issue. It would be great if someone among the dbus maintainers could act as sponsor. > I do not expect that dbus-broker will be compatible with every D-Bus > service in Debian. The one incompatibility that I'm reasonably sure exists > is that if the Exec= for an activatable service points to a command that > will fork (background itself) and exit 0, dbus-daemon tolerates this > (at the cost of worse error behaviour because it cannot tell whether > the service subsequently fails), while dbus-broker almost certainly does > not. This is inadvisable behaviour even with the reference dbus-daemon, > so I'd consider it to be a bug in the service, but unfortunately it can't > be detected statically. That would need to be investigated indeed. So far the package works well on a couple of very minimal installs I tried it on. Thanks! Cheers, Dan
