Fatal warnings in client code

[Thiago Macieira]

On Monday, 6 March 2023 14:33:15 PST Thomas Kluyver wrote:
> I think this really depends what the application is doing with D-Bus. If
> it's using D-Bus solely to send a desktop notification, it probably doesn't
> require all that much care to cope with disconnection, and users would
> probably rather just the notification was lacking, instead of the entire
> application dying (especially if they had some unsaved work in that
> application). I appreciate that a hard crash is easier to debug, though.

Yes, for status notification, it looks like it could just ignore the fact that 
the bus connection went away. It's wasting resources, but can probably survive 
without other ill effects.

Other D-Bus protocols or implementations thereof may not be as forgiving. 
Suppose that the application is trying to talk to bluez or udisks or 
PulseAudio or whatever else: is the application prepared to cope with the fact 
that it's going to get an error for the method call it's sending?

Even for StatusNotifier, is the code in question ready for the fact that it 
can't send the signal messages?

-- 
Thiago Macieira - thiago (AT) macieira.info - thiago (AT) kde.org
   Software Architect - Intel DCAI Cloud Engineering