Patch to support multiple devices in NM 0.7
christian.huff at silur.de
Thu May 7 04:51:10 EDT 2009
Ka-Hing Cheung schrieb:
> On Wed, 2009-05-06 at 21:30 +0200, Christian Huff wrote:
>> 4 days ago, I uploaded the 5th version of my patch to support multiple
>> devices in libpurple's NetworkManager integration. I found pidgin not
>> only to stall if you discoonect a device that might have been critial to
>> network connectivity but also if you connect one.
>> With the 5th version of the patch, libpurple will always disconnect and
>> connect if there is a change in the number of devices that are online. I
>> figured this scheme to be most reliable. Ultimately, this scheme makes
>> libpurple behave as if it was talking to an earlier version of
>> NetworkManager, where there was only one active device at any time and a
>> change of the active device would always result in reconnect actions by
> This will make me very unhappy. When I am at work and my laptop is
> typically on both ethernet and wireless at the same time, because
> sometimes I unplug and walk around. Sometimes wireless acts a little
> funny and it will suck a lot to sign on and off because of that.
I understand, but haven't you experienced the problem with pidgin
stalling? The patch should fix you needing to manually reconnect if you
disconnect or reconnect a device. The problem is that if you had
previously been on wireless, and add the ethernet, at least with Ubuntu,
there is a NetworkManager policy that forces eth0 default for DNS and
routes. This leaves me unable to receive messages as well, so it's not
limited to devices going down. I can reproduce that.
I'm always looking for ways to improve, maybe a
reconnect-on-every-device-up, reconnect-on-critical-device-down scheme
works better. I'll give that a try. It wouldn't fix your problem of
funny wireless behaviour (like it's going up and immediately down), as
it would only prevent the reconnect on going down. So far, I am unable
to determine properly if a device that is going up will or will not
become part of a default active connection, thus creating the need for a
reconnect. I'm hoping for a reply from the NetworkManager devs, however.
khc, are you willing to test the patch?
More information about the Devel