Lin:
> Thanks Brian, I will rework it.
I took care of it. I was working on fixing the problem in the
appearance capplet, which is also in the SUNWdesktop-prefs package, so
I went ahead and implemented the fix. I tested it and it works also.
I will go ahead and file a bug report to get this upstream
Thanks for helping to fix this.
Brian
> I have one question. Why some applications go to /usr/lib? And seems not
> all of them are daemons. So is there a policy tell us which type of
> applications should go to /usr/lib?
All daemons (or programs that users are not normally to run by hand),
should be installed to "libexec", which translates to /usr/lib on
Solaris.
I think the one remaining daemon that we really should move to libexec
is gnome-keyring-daemon. Seahorse also has a seahorse-daemon which
we should move also when we integrate it.
Brian
> Brian Cameron wrote:
>> Lin:
>>
>>> g-s-d invokes g_spawn_async to search esd in G_SPAWN_SEARCH_PATH which
>>> is the user's PATH. Since esd is in /usr/lib in Solaris, so hard code it
>>> as a branding.
>> Great work tracking down this problem. However, this patch cannot go
>> upstream since hard-coding the path to "/usr/lib" is not appropriate.
>>
>> Note that the /usr/lib/esound.pc file specifies the location of the
>> daemon. If you run this command, it tells you the directory where the
>> ESD daemon is installed:
>>
>> % pkg-config --variable=esd_serverdir esound
>>
>> Note the patch dbus-glib-01-findlaunch.diff which uses this same
>> technique to find the location of the dbus-daemon. I would recommend
>> that you rework this patch so that it works similarly to the
>> dbus-glib-01-findlaunch.diff patch. Note how this patch adds the
>> trailing "/" in the configure.ac file, so that if building with an
>> older version of dbus which doesn't support the "daemondir" variable
>> that it will still work. I'd recommend using the same technique in
>> case someone tries building g-s-d with an older version of esd which
>> may not support the "esd_serverdir" variable. In other words, I'd
>> add something like this to configure.in
>>
>> +GSD_ESD_LOCATION=`pkg-config esound --variable=esd_serverdir`
>> +if test x"$GSD_ESD_LOCATION" != "x"; then
>> + # Add slash, so that if the ESD pkgconfig file does not
>> + # specify esd_serverdir, the tools/Makefile.am file will look
>> + # in the PATH for the daemon, as was previous behavior.
>> + #
>> + GSD_ESD_LOCATION="$DBUS_DAEMONDIR/"
>> +fi
>> +AC_SUBST(GSD_ESD_LOCATION)
>>
>> Note that the AC_SUBST macro only modifies the Makefiles, so you
>> probably want to add this to the plugins/sound/Makefile.am file which
>> builds the gsd-sound-manager.c file:
>>
>> libsound_la_CFLAGS = \
>> + -DGSD_ESD_LOCATION=\"$(GSD_ESD_LOCATION)\" \
>> $(SETTINGS_PLUGIN_CFLAGS) \
>> $(GNOME_CFLAGS) \
>> $(ESD_CFLAGS) \
>> $(AM_CFLAGS)
>>
>> Then you could change gsd-sound-manager.c like this:
>>
>> + char *argv[] = { GSD_ESD_LOCATION "esd", "-nobeeps", NULL};
>>
>> Also I would leave the G_SPAWN_SEARCH_PATH in the g_spawn_asynch line.
>> Then it should still work okay even if GSD_ESD_LOCATION ends up being ""
>> if using an older version of esound. Using G_SPAWN_SEARCH_PATH should
>> work okay even if you pass in the full path.
>>
>> Then the patch could go upstream:
>>
>> Brian
>
>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: control-center.diff
URL:
<http://mail.opensolaris.org/pipermail/jds-review/attachments/20080506/281f4e11/attachment.ksh>
