Re: [PATCHES] fixes for some smaller issues

[Dan Williams <dcbw redhat com>]

On Thu, 2008-01-10 at 02:53 +0100, Michael Biebl wrote:
> Dan Williams wrote:
> > On Sun, 2008-01-06 at 17:37 +0100, Michael Biebl wrote:
> > 
> >> 3.) man_page_section.patch
> >> After applying 2.), this patch fixes the sections (1->8) of 
> >> NetworkManager(Dispatcher) (already in trunk).
> >> Please also do a
> >>
> >> cd man
> >> svn mv NetworkManager.1.in NetworkManager.8.in
> >> svn mv NetworkManagerDispatcher.1.in NetworkManagerDispatcher.8.in
> > 
> > Committed.
> 
> Forgot to update the sections/references in the man pages itself.
> 
> Please apply attached man_section.patch too.
> 
> > 
> >> 4.) no_gnome_disable_deprecated.patch
> >> That's required to allow nm-vpn-properties to compile with gtk-2.12.
> >> (already in trunk network-manager-applet)
> > 
> > Committed.
> > 
> >> Remains the most controversial patch
> >> 5.) rfkill.patch
> >> I already brought this topic up: hal changed the signature killswitch 
> >> dbus arguments from uint to int in hal_0.5.10.
> >> You modified my original patch to first try with uint, then with int 
> >> (commit 3160). Unfortunately this doesn't work, as dbus will issue an 
> >> assert with hal_0.5.10 as the status int variable has the wrong type.
> >> We could try to mess around with two status variables (int and uint), 
> >> but this is ugly imho. As we already require a very recent libnl, which 
> >> only up-to-date distros ship, I simply remove the support for hal < 
> >> 0.5.10 and bumped the build dependencies in configure.in accordingly.
> > 
> > I committed r3226 in which I took your suggestion to have two different
> > variables for the different calls.  Could you test?  I haven't
> > run-tested it yet but it looks correct to me (though I've been wrong
> > before :).
> 
> Hm, I still get a lot of error messages:
> NetworkManager: <info>  Wireless now enabled by radio killswitch
> process 12045: arguments to dbus_message_get_args() were incorrect, 
> assertion "(error) == NULL || !dbus_error_is_set ((error))" failed in 
> file dbus-message.c line 1667.
> This is normally a bug in some application using the D-Bus library.
> process 12045: arguments to dbus_message_get_args() were incorrect, 
> assertion "(error) == NULL || !dbus_error_is_set ((error))" failed in 
> file dbus-message.c line 1667.

Looks like we need some dbus_error_init() calls there, seems like I
forgot those in the initial bits anyways.

Dan


> This is normally a bug in some application using the D-Bus library.
> NetworkManager: <info>  Error getting killswitch power arguments: 0� - 
> (null)
> ....
> 
> I have hal 0.5.10 btw.
> 
> What about the attached patch hal_killswitch.patch instead, which adds a 
> configure check and some ifdefs. I tested it with hal 0.5.10 and It's 
> working fine (and should also with hal < 0.5.10, but please test).
> It has the advantage, that it get's rid of the error messages and you 
> don't have to make 2 dbus calls for hal 0.5.10. The downside is, that 
> the package has to be recompiled for a hal 0.5.9 -> hal 0.5.10 upgrade.
> 
> Cheers,
> Michael
> 
> P.S:
> What about the Makefile cleanups for trunk I sent in my previous emails?
> I also prepared one for the stable branch. See attached built_sources.patch.
>