Python bindings ready for first commit

Seth Nickell
24 Sep 2003 22:51:22 -0700

On Mon, 2003-09-22 at 09:38, Havoc Pennington wrote:
> On Mon, 2003-09-22 at 01:45, Seth Nickell wrote:
> > Well, I was getting ready to commit this and did a "cvs update" and
> > everything seems to have broken (this is on the dbus-object-names
> > branch). dbus_bus_acquire_service() now returns the error:
> > 
> > A security policy in place prevents this sender from sending this
> > message to this recipient, see message bus configuration file (rejected
> > message had interface "org.freedesktop.DBus" member "AcquireService"
> > error name "(unset)" destination "org.freedesktop.DBus")
> > 
> > I can't see anything that would cause this in my session.conf file... so
> > I'm wondering if one of today's checkins caused the problem.
> > 
> I probably broke it, if you can write a regression test for the failure
> you see that will let me fix it in about 10 minutes, plus we'll have a
> regression test. ;-) Do you see an easy way to test it?

I'm not quite sure where to write a regression test for this: its
totally contingent on your config files and hence what daemon is
running. This does get got in the tests for me: If you have a daemon
running already and run "bus-test", it does fail for me (starting with
the call to AcquireService). The basic problem is that the allow all
that should be active on the session bus isn't "taking".

If I add this:
    <allow send_destination="org.freedesktop.DBus"/>
    <allow receive_sender="org.freedesktop.DBus"/>

To session.conf file... I can make calls to org.freedesktop.DBus
(including AcquireService).

Christophe Fergeau (who also sees this) thinks it was caused by the "Get
matching rules mostly working in the bus" commit. Unfortunately a little
pressed for time as of late, but I'll look into debugging more deeply if
the problem is still extant this weekend.