Re: Using a mobile phone via Bluetooth

[Dan Williams <dcbw redhat com>]

On Tue, 2009-07-07 at 17:43 +0100, Brian Morrison wrote:
> Dan Williams wrote:
> 
> > Sounds like more on the Bluetooth side, of course.  Does your phone do
> > anything when NM tries to bring the connection up?  Mine usually asks me
> > whether I want to allow the computer to connect, or it'll show an icon
> > saying a computer is connected.
> 
> Bluetooth will insist on being paired and authorised as well or it will
> ask for permission to allow the connection and use it.

Right, you can't always just open the persistent rfcomm port and expect
it to work, due to these sorts of issues.

> >  In any case, this is exactly why this
> > method is a hack, and why we want the real solution...
> 
> I don't see there is any way to bypass these parts of the Bluetooth
> protocol stack, but perhaps you know differently.

Well, the problem is that there's no intelligence built into that hack
method of doing Bluetooth.  In the bits we've added to 0.8 for
Bluetooth, we ask Bluez to set up the rfcomm port, and we can get
reliable errors when we can't talk to the phone or when something else
goes wrong.  With the persistent stuff, the errors just get completely
lost, because to NetworkManager, the hacked rfcomm thing just looks like
a directly connected serial port and thus all errors on the bluetooth
side just end up looking like a generic failure to talk to the modem.

Dan