logoff on suspend

[Adam Nowotny <adam@…>]

Gajim should log off before the system is suspended, which is a common event in a mobile environment. I've used this feature with Psi, and it was very helpful. Right now i get these annoying popups every time turn the laptop back on. This is in progress in #859, but logoff is helpful for people subscribed to my presence. The problem is portability here, it's possible in Windows, but have no idea how's it with other OSes.

[dottedmag@…]

What about using gajim-remote in your suspend/resume scripts? Seems to be

trivial task.

[Adam Nowotny <adam@…>]

I using windows, no script there i guess. Besides i'm the user here, not

developer.

[nk]

anyone that knows can link us to Psi source (I think they do this stuff)

and they do it cross plat?

[jt]

Unfortunately, this does not work, at least not on my debian system. I put

"su -c 'gajim-remote change_status offline' $USER" in my suspend script. However, the script runs in a context that does not allow using dbus. At least I could not get it to work. Any suggestions?

[dkirov]

This is what are signals for. Gajim should disconnect all accounts from

servers on TERM signal and you will just do: kill -15 pid_of_gajim in your suspend script.

[sb]

Gajim quits on TERM signal. I think it is better to have it disconnect on suspend and connect again on wake up. This is something that is better done in conncection with gnome-power-manager since this allows a generic implementation across distributions, compared to acpi scripts.

[sb]

It should be alright for mobile entvironment now if network manager is used, at least on linux.

[patrys]

Not if NetworkManager? does not support your distro (which is the case here - PLD Linux).

[nk]

MACOSX/Windows http://svn.psi-im.org/listing.php?repname=Psi&path=%2Ftrunk%2Fsrc%2Ftools%2Fsystemwatch%2F#_trunk_src_tools_systemwatch_

[2:12:53 μμ] spike (psi dev): all 3 logoff i would expect
[2:12:56 μμ] nkour: okay
[2:12:58 μμ] spike (psi dev): well, shutdown that is for sure
[2:13:08 μμ] spike (psi dev): suspend and hibernate on windows, both logoff
[2:13:10 μμ] nkour: any idea where you have this code?
[2:13:19 μμ] nkour: which file I should look @?
[2:13:54 μμ] nkour: I guess you also save show, so when you come back from suspend you go that show
[2:14:06 μμ] nkour: if you shutdown, then I don't think you use that show anymore
[2:14:07 μμ] spike (psi dev): well, we don't do that yet, but we should do that yes
[2:14:16 μμ] nkour: oh ok
[2:14:18 μμ] spike (psi dev): we should remember the last available status before shutting down
[2:14:18 μμ] nkour: which file is it?
[2:14:40 μμ] nkour: it's interesting how you do that in MacOSX and in Windows (maybe even Vista works)
[2:14:45 μμ] spike (psi dev): http://svn.psi-im.org/listing.php?repname=Psi&path=%2Ftrunk%2Fsrc%2Ftools%2Fsystemwatch%2F#_trunk_src_tools_systemwatch_
[2:14:47 μμ] spike (psi dev): systemwatch

for Unix:

[2:11:46 μμ] spike (psi dev): it uses SIGUSR, which sounds like something non-standard
[2:11:52 μμ] spike (psi dev): we had a patch that does it on unix

[nk]

we should watch for:

// These defines come from Microsoft Platform SDK, August 2005
#if(WINVER >= 0x0400)
# ifndef WM_POWERBROADCAST
#  define WM_POWERBROADCAST                0x0218
# endif // WM_POWERBROADCAST
# ifndef _WIN32_WCE
#  ifndef PBT_APMQUERYSUSPEND
#   define PBT_APMQUERYSUSPEND             0x0000
#  endif // PBT_APMQUERYSUSPEND
#  ifndef PBT_APMQUERYSTANDBY
#   define PBT_APMQUERYSTANDBY             0x0001
#  endif // PBT_APMQUERYSTANDBY
#  ifndef PBT_APMQUERYSUSPENDFAILED
#   define PBT_APMQUERYSUSPENDFAILED       0x0002
#  endif // PBT_APMQUERYSUSPENDFAILED
#  ifndef PBT_APMQUERYSTANDBYFAILED
#   define PBT_APMQUERYSTANDBYFAILED       0x0003
#  endif // PBT_APMQUERYSTANDBYFAILED
#  ifndef PBT_APMSUSPEND
#   define PBT_APMSUSPEND                  0x0004
#  endif // PBT_APMSUSPEND
#  ifndef PBT_APMSTANDBY
#   define PBT_APMSTANDBY                  0x0005
#  endif // PBT_APMSTANDBY
#  ifndef PBT_APMRESUMECRITICAL
#   define  PBT_APMRESUMECRITICAL          0x0006
#  endif // PBT_APMRESUMECRITICAL
#  ifndef PBT_APMRESUMESUSPEND
#   define PBT_APMRESUMESUSPEND            0x0007
#  endif // PBT_APMRESUMESUSPEND
#  ifndef PBT_APMRESUMESTANDBY
#   define PBT_APMRESUMESTANDBY            0x0008
#  endif // PBT_APMRESUMESTANDBY
#  ifndef PBTF_APMRESUMEFROMFAILURE
#   define PBTF_APMRESUMEFROMFAILURE       0x00000001
#  endif // PBTF_APMRESUMEFROMFAILURE
#  ifndef PBT_APMBATTERYLOW
#   define PBT_APMBATTERYLOW               0x0009
#  endif // PBT_APMBATTERYLOW
#  ifndef PBT_APMPOWERSTATUSCHANGE
#   define PBT_APMPOWERSTATUSCHANGE        0x000A
#  endif // PBT_APMPOWERSTATUSCHANGE
#  ifndef PBT_APMOEMEVENT
#   define PBT_APMOEMEVENT                 0x000B
#  endif // PBT_APMOEMEVENT
#  ifndef PBT_APMRESUMEAUTOMATIC
#   define PBT_APMRESUMEAUTOMATIC          0x0012
#  endif // PBT_APMRESUMEAUTOMATIC
# endif // _WIN32_WCE
#endif // WINVER >= 0x0400

[nk]

I looked everywhere. all I could find was win32event.WaitForSingleObject?

and yeah I don't feel like it atm

[nk]

via WMI (should be simpler):

http://msdn.microsoft.com/library/default.asp?url=/library/en-us/wmisdk/wmi/win32_powermanagementevent.asp

Must use this file: http://tgolden.sc.sabren.com/python/wmi.html to access to WMI

[asterix]

I did it using wmi, but the signal is emitted too late. Gajim doesn't have time to go offline before going to suspend :/

[Kamil Páral]

I second this bug. When I shutdown my notebook, Gajim correctly logs off. But if I suspend or hibernate, I am "online" for another 5-15 minutes. All messages which comes in the meantime are lost. That's very inconvenient.

Gajim should listen for suspend/hibernate signals and log off, and log in again after resume.

I use Ubuntu 8.04.