temporary suspending providers

[Bastien Nocera]

On Mon, 2010-05-10 at 13:21 +0200, Guilhem Bonnefille wrote:
> Fun: I think about this topic this week-end.
> 
> The major question is: is Geoclue responsible of hardware, or only
> aware of software.
> 
> IMHO, Geoclue must keep simple and only responsible of software layer.
> Concerning the hardware layer, Geoclue must handle correctly the
> disappearing of hardware but should not be actor of it. It is the
> responsibility of an other layer to drive hardware configuration
> (something like a part of FSO).
> 
> The idea is to keep each layer simple and cooperative. By this way, we
> will be able to build complex solutions with Geoclue on environment
> where harware is persistent (like desktop) or where hardware can be
> enable/disable (like mobile device).
> 
> For exemple, the provider associated with GPS device should declare
> itself as unavailable when gps device is disabled.

That shouldn't really matter to Geoclue, if the device isn't there. The
problem is disabling hardware when there's no listeners for that
particular type of device.

So if geoclue-master only has one client, and requested us not using the
GPS, we shouldn't have the GPS providers, or gypsy, started.

And gypsy should shut itself down if the last of the devices it was
handling got removed.

Both are already opened as bugs against Geoclue and Gypsy, I believe,
it's just a matter of somebody implementing the code.

I might look at it in the future, but it will probably be as part of a
port to gdbus, and API rework.

Cheers