(pruninc cc list since this is a regular pulseaudio bug) On Fri, Dec 8, 2017 at 12:34 PM, Vincent Lefevre <vinc...@vinc17.net> wrote: > On 2017-12-08 11:22:02 -0300, Felipe Sateler wrote: >> On Thu, Dec 7, 2017 at 10:03 PM, Vincent Lefevre <vinc...@vinc17.net> wrote: >> > On 2017-12-07 18:08:47 +0000, Simon McVittie wrote: >> >> The recommendation should ideally be something like >> >> "dbus-user-session | NOT(booted using systemd)", but we don't have the >> >> syntax to express that. >> > >> > How about a new package pulseaudio-no-systemd and a Recommends on >> > "dbus-user-session | pulseaudio-no-systemd"? This package would >> > do almost nothing. It could possibly change the configuration to >> > re-enable autospawn. >> >> It would be weird to have two snippets installed, one that enables and >> the other disables pulseaudio. > > Yes. But shouldn't the default be no-autospawn? Then the > /etc/pulse/client.conf.d/00-disable-autospawn.conf config file > would no longer be needed, and pulseaudio-no-systemd would just > add a config file to enable autospawn.
This may work. Could you propose a patch for this scheme? An old patch of mine that was ultimately not accepted upstream could serve as a starting point: https://lists.freedesktop.org/archives/pulseaudio-discuss/2014-November/022696.html (Remember we need to keep autospawn by default on for nonlinux archs. > >> But I don't want to make it unnecessarily difficult either. I'm open >> to suggestions to improve the current situation, but they must >> fulfill the following properties: >> >> 1. It should use the systemd --user manager by default >> 1b. We should keep a single place to disable, which would be the >> systemd --user manager, and thus autospawn needs to be disabled by >> default. >> 2. This configuration should not be changed unless explicitly chosen >> by the admin/user. > > pulseaudio-no-systemd could also just ask the user whether he wants > to enable autospawn. > > But you could also see autospawn as being the default configuration > for sysvinit (where pulseaudio-no-systemd would be installed by > default). Thus it would not really be a change. > > Alternatively, couldn't the PulseAudio client library be smarter, > i.e. be able to know whether it should autospawn or not by default, > without the need of a configuration file? This is a bit too much magic for my taste. I'm not willing to carry a patch like this if it is not accepted upstream first. -- Saludos, Felipe Sateler
