Please signoff for both architecture. This release needs a bit testing,
as upstream changed X11 autolaunch support. I re-enabled this with a
hack in dbus-core instead of makedepending on libx11.

On Mon, May 9, 2011 at 4:56 PM, Jan de Groot <jan@jgc.homeip.net> wrote:
> Please signoff for both architecture. This release needs a bit testing,
> as upstream changed X11 autolaunch support. I re-enabled this with a
> hack in dbus-core instead of makedepending on libx11.

Possibly stupid question: if dbus-core still compiles with your hack,
doesn't that mean that dbus should not need libx11 as a makedepeng in
the first place (i.e. requiring it is an upstream bug)?

Cheers,

Tom

On Mon, 2011-05-09 at 17:33 +0200, Tom Gundersen wrote:
> On Mon, May 9, 2011 at 4:56 PM, Jan de Groot <jan@jgc.homeip.net> wrote:
> > Please signoff for both architecture. This release needs a bit testing,
> > as upstream changed X11 autolaunch support. I re-enabled this with a
> > hack in dbus-core instead of makedepending on libx11.
>
> Possibly stupid question: if dbus-core still compiles with your hack,
> doesn't that mean that dbus should not need libx11 as a makedepeng in
> the first place (i.e. requiring it is an upstream bug)?

In our previous version, libdbus would auto-launch a session bus trough
some X11 cookie stuff. If dbus was compiled without X, then dbus-launch
would not succeed, so compiling that code without support for X is
useless. They ifdef'ed that code in libdbus.
Now with our package, we build dbus-core as a package without
dbus-launch and replace the dbus-launch binary with the X11-enabled
version in the dbus package. This will only work if the conditional code
in libdbus is enabled, which can either be done by makedepending on
libx11, or by sed'ing the conditional in the associated file. I chose to
do the 2nd.

On Mon, May 9, 2011 at 8:27 PM, Jan de Groot <jan@jgc.homeip.net> wrote:
> On Mon, 2011-05-09 at 17:33 +0200, Tom Gundersen wrote:
>> On Mon, May 9, 2011 at 4:56 PM, Jan de Groot <jan@jgc.homeip.net> wrote:
>> > Please signoff for both architecture. This release needs a bit testing,
>> > as upstream changed X11 autolaunch support. I re-enabled this with a
>> > hack in dbus-core instead of makedepending on libx11.
>>
>> Possibly stupid question: if dbus-core still compiles with your hack,
>> doesn't that mean that dbus should not need libx11 as a makedepeng in
>> the first place (i.e. requiring it is an upstream bug)?
>
> In our previous version, libdbus would auto-launch a session bus trough
> some X11 cookie stuff. If dbus was compiled without X, then dbus-launch
> would not succeed, so compiling that code without support for X is
> useless. They ifdef'ed that code in libdbus.
> Now with our package, we build dbus-core as a package without
> dbus-launch and replace the dbus-launch binary with the X11-enabled
> version in the dbus package. This will only work if the conditional code
> in libdbus is enabled, which can either be done by makedepending on
> libx11, or by sed'ing the conditional in the associated file. I chose to
> do the 2nd.

Thanks for the explanation, now I get it.

I didn't notice any problems on my laptop (using KDE), so: signoff x86_64.

Cheers,

Tom

[2011-05-09 16:56:45 +0200] Jan de Groot:
> Please signoff for both architecture. This release needs a bit testing,
> as upstream changed X11 autolaunch support. I re-enabled this with a
> hack in dbus-core instead of makedepending on libx11.

Signoff x86_64.

--
Gaetan