> > > > +static void set_delay_report(const GDBusPropertyTable *property,
> > > > + DBusMessageIter *iter, GDBusPendingPropertySet id,
> > > > + void *data)
> > > > +{
> > > > + struct media_transport *transport = data;
> > > > + uint16_t arg;
> > > > + int err;
> > >
> > > This really needs to check that sender is the owner of the transport.
> >
> > The problem is that currently there is no mechanism which will allow
> > to get the sender name in the property get/set callback (at least I
> > couldn't find any). The statement "this property is only writable when
> > the transport was acquired by the sender" is also for the Volume
> > property, but as for Delay (which currently lacks the setter) it's not
> > implemented. So, now the question is whether you would like to fix
> > that as well (or amend the doc)? Some users might say that allowing
> > others to control volume might be a feature. For the Delay I'm not
> > sure... Probably you are right that the Delay is more intrinsic to the
> > transport, and external manipulation would not be desired.
> >
> > As for the design of this authorization check I can see two possibilities:
> >
> > 1. Pass `message` to the `property->set()` in the
> > "dbus/objects.c:properties_set()". But this will require updates in
> > all setters (and maybe for symmetry the getter should also receive the
> > original message?).
> > 2. Add a dedicated callback for setter/getter authorization, in a
> > similar way the `exists()` works.
> >
> > In either way I think that this should be a separate patch, applied
> > either before or after the Delay work.
>
> Right, or we could just implement something like get_sender_by_id then
> the callback can just call it to check who is the sender, that way we
> don't need to modify existing code if it doesn't care about it.
Yeap, that was my 3rd idea, but I thought that it would be a bit
hackish since the `pending_property_set` seems to be kinda private.
But OK, I can make it that way. It would be definitely the easiest
way.
