(original) (raw)

On Dec 8, 2009, at 9:43 AM, Tobias Arrskog <topfs2@xboxmediacenter.com> wrote:

To bring more to the discsussion regarding uses of events at both wakeup and suspends.

On suspend apps can close sockets and signal disconnects properly to servers (even if the server in itself probably should handle a wrong disconnect it's nice to support it nicely).

Hmmm... I know that my common suspend use case involves closing the lid and putting my laptop in my backpack, so I want suspend to happen as quickly and reliably as possible, and waiting for network connections to cleanly shut down strikes me as creating another set of failure points. I would rather rely on userspace network apps cleaning up after themselves at resume if needed.

On resume sockets could get reopened.
In XBMC, LIRC comes to mind as a perfect candidate, it tends to need reinitiate and we want suspend to behave transparently as shutdown, so resume needs to look like a startup so we rescan and such.

On Tue, Dec 8, 2009 at 4:36 PM, Tobias Arrskog <topfs2@xboxmediacenter.com> wrote:
Sure I agree, I just made it quick (more as a proof of concept for a patch I was working on).

But glad to see that there is more than me that needs it atleast :D

On Tue, Dec 8, 2009 at 4:34 PM, Richard Hughes <hughsient@gmail.com> wrote:

2009/12/8 Tobias Arrskog <topfs2@xboxmediacenter.com>:

> I posted a patch a few weeks ago, It might have been overlocked or just
> uggly :)
> Basically it did a dbus-event from pm-utils.
> http://lists.freedesktop.org/archives/devkit-devel/2009-November/000527.html

No, you need to define the signal in the DeviceKit-power introspection
document and emit the signal from within DKp. I think doing this as
pm-utils level is the wrong layer.

Richard.