Re: [Evolution-hackers] Wrong factories starting after alternative installs?



On Mon, 2010-11-29 at 17:55 -0500, Paul Smith wrote:
> That's exactly where I got my version originally.  I did, just to be
> sure, try copying that verbatim into session.local.conf then logging
> out and back in, but it didn't help.

OK, I read some man pages about dbus-daemon and I did some
experimentation with strace -p attached to my session daemon.  I
discovered that if you run kill -HUP your session daemon, it WILL reread
your config files.  Yay!  Easier than logging out/back in.

When I do that and look at what files are opened containing
"AddressBook" I see that first it opens the system service file and read
that:

open("/usr/share/dbus-1/services/org.gnome.evolution.dataserver.AddressBook.service", O_RDONLY) = 84

then afterwards it opens my alternate service file and reads that:

open("/opt/evo-2.32/share/dbus-1/services/org.gnome.evolution.dataserver.AddressBook.service", O_RDONLY) = 84

Then it ends up using the path it found in the FIRST file.  According to
the dbus-daemon man page it's SUPPOSED to work like this:

        <servicedir>
        Adds a directory to scan for .service files. Directories are
        scanned starting with the last to appear in the config file (the
        first .service file found that provides a particular service
        will be used).

OK.  So, it's definitely using the first .service file, as the
parenthetical statement says.  However, when I examine the contents
of /etc/dbus-1/session.conf I cannot figure out how it can possibly be
looking at /usr/share/dbus-1 first, since it says "directories are
scanned starting with the last to appear in the config file"... as far
as I can see my stuff comes last.


However, if I rename the system .service file to something else, then my
alternate service DOES get chosen and run and everything works, so I'm
sure that it's not some syntax error in my part but rather some kind of
ordering issue that I don't understand.


Sigh.  Everyone is happy to add all kinds of funky new features, and
their added complexity, but no one ever thinks about
maintainability/debug-ability... until it's too late.



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]