ke, 2010-01-13 kello 22:07 +0100, Lennart Poettering kirjoitti:
> On Wed, 13.01.10 07:35, Tanu Kaskinen (tanuk at iki.fi) wrote:
> > This is how it might look, if D-Bus settings are added (the actual
> > syntax is subject to discussion, I haven't thought hard what kind of
> > syntax would be the best):
> >
> >
> >{ecstasy}unix:[native]/tmp/pulse-6f7zfg/native,[dbus]/tmp/pulse-6f7zfg/dbus
> >tcp6:ecstasy.ring2.lan:[native]4713,[dbus]24883
> >tcp:ecstasy.ring2.lan:[native]4713,[dbus]24883
>
> Kinda makes sense, though maybe use different prefixes altogether for
> that? i.e:
>
> tcp4:, tcp6:, and unix: continue to mean "native protocol over the
> resp lower level transport. And then we add dbus-tcp4:, dbus-tcp6:,
> dbus-unix: and so on. And a client looking for a server to connect to
> then goes through that list and picks the first entry that works for
> it that speaks the right protocols.
>
> Using just different prefix seems like a simpler, more compatible
> extension to the current syntax.
Simpler, yes. More compatible, not much. The only cases that I can see
the proposed syntax would break are socket paths that
* start with "[" (ie. they are relative paths, which don't make any
sense anyway, I think) or
* contain commas.
> Does that make sense to you?
Well, the simplicity point is valid. But if the user wants to set the
server manually, completely separate server strings for different
pulseaudio protocols require him to always specify hosts twice, which is
annoying and a bit error prone. Based on very little empirical data, I
would guess that the most common case for manual server configuration is
when just the server hostname is needed. I would like this to still
work:
$ PULSE_SERVER=gurki pavucontrol
(Ok, that would still work, because pavucontrol doesn't currently use
D-Bus, but you get the idea.)
--
Tanu Kaskinen
