On Fri, 29 Feb 2008 11:52:18 +0000
Darren Kenny <Darren.Kenny at Sun.COM> wrote:
> Actually, another thing is the use of dbus-launch - why are you doing this?
>
>
> The command:
>
> su darrenk -c "env HOME=/home/darrenk DISPLAY=:0 notify-send test"
>
> works for me, but
>
> su darrenk -c "env HOME=/home/darrenk DISPLAY=:0 dbus-launch
> notify-send test"
>
> causes several dbus-daemon processes to remain hanging about.
>
> Maybe that's what the ctrun helps with, but it still seems like an unnecessary
> overhead. Unless the is an edge case I've not seen here...
If you can figure out a better way I'm all ears.
Some history.
In the current daemon you will see that on exit we pkill dhcpagent on
exit. This is because when we do a 'ifconfig dhcp' that starts
dhcpagent which ends up in our contract but it doesn't exit when we do
so we end up hanging, timing out, and ending up in maintenance.
I see the same thing with notify-send in some cases. The typical case
is on startup notify-send uses 'dbus-launch --autolaunch' which then
starts up the session bus and starts notify-send leaving us in much the
same situation as with ifconfig. ctrun won't exit 'cuz there are still
parts of the contract hanging around. If instead we have ctrun <-
dbus-launch <- notify-send then it dies off notify-send, dbus-launch,
ctrun and all is well (I think).
The problem with cli testing is that you already have the
DBUS_SESSION_BUS_ADDRESS setup when you start notify-send so you don't
see the extra autolaunch step. Unset that and then fiddle around.
nwamd doesn't have it set 'cuz we are started by startd well before
gnome (or at least outside the environment of gnome).
mph
>
> Darren.
>
> Darren Kenny wrote:
> > Hi Michael,
> >
> > A comment about how you call notify-send.
> >
> > You don't need to include the full path to the icons, infact it's better if
> > you
> > don't since then it will use the appropriate icon based on the theme.
> >
> > e.g.
> >
> > notify-send --icon connect_no "test"
> > notify-send --icon connect_creating "test"
> > notify-send --icon connect_established "test"
> >
> > Darren.
> >
> > Michael Hunter wrote:
> >> On Fri, 29 Feb 2008 10:54:57 +0000
> >> Alan Maguire <Alan.Maguire at Sun.COM> wrote:
> >>
> >>> hi Michael
> >>>
> >>> found one additonal issue - in testing the BFU
> >>> image, i'm not seeing the notify-send dialogs, and a look
> >>> at the nwamd SMF log reveals that ctrun is failing:
> >>>
> >>> ctrun: invalid lifetime ' child'
> >>> Usage: ctrun [-i eventlist] [-f eventlist] [-l lifetime]
> >>> [-o optionlist] [-r count [-t]] [-v]
> >>> [-F fmri] [-A aux] command
> >>>
> >>> (note the leading space in ' child'). i suspect
> >>> splitting the "-l child" string parameter to
> >>> start_child() out into "-l", "child" (line 126 of
> >>> interface.c) may fix this.
> >> When I was cleaning up the code in prep for final approach I somehow
> >> combined these. In any case I fixed it earlier but didn't update the
> >> webrev. I just updated the webrev and the bfu archive is correct.
> >>
> >> mph
> >>
> >>> Alan
> >>>
> >>> Michael Hunter wrote:
> >>>> I have a webrev which is intended to solve:
> >>>>
> >>>> 6635801 CONN_DEC_REF() uses <= to compare a uint32_t with 0
> >>>> 6642066 NWAM Phase 0 needs better observability
> >>>>
> >>>> at http://cr.opensolaris.org/~mph/phase0_notify/
> >>>>
> >>>> Alan and Sebastien have reviewed this while it was being written. If
> >>>> anybody else has a chance to look at it soon any comments would be
> >>>> appreciated.
> >>>>
> >>>> mph
> >>>> _______________________________________________
> >>>> nwam-discuss mailing list
> >>>> nwam-discuss at opensolaris.org
> >>>> http://mail.opensolaris.org/mailman/listinfo/nwam-discuss
> >>>>
> >> _______________________________________________
> >> nwam-discuss mailing list
> >> nwam-discuss at opensolaris.org
> >> http://mail.opensolaris.org/mailman/listinfo/nwam-discuss
> > _______________________________________________
> > nwam-discuss mailing list
> > nwam-discuss at opensolaris.org
> > http://mail.opensolaris.org/mailman/listinfo/nwam-discuss
