On Sat, 07.05.11 00:31, Ryan Lortie ([email protected]) wrote: > > hi Sander, > > On Fri, 2011-05-06 at 16:58 -0500, Sander Jansen wrote: > > Why? For what purpose? > > There are quite a lot of reasons that I can imagine why this might be > useful. > > For example, desktop environments could use it to match between the DBus > name acquired by the application (which would be the same as the > application ID) in order to match it with the desktop file.
Hmm, I think this deserves maybe a spec of its own, i.e. a definition what an app id is, and its format (i.e. along the line of dbus bus names), plus where it should be used (bus names, .desktop file names, XDG_RUNTIME_DIR subdirs, ~/.config/ and so on). If you really plan to go for this then this would probably be a major undertaking, and we really should mandate this all the other XDG specs too. i.e. in the .desktop spec, in the XDG basedir spec (i.e. regards subdirs in ~/.config and XDG_RUNTIME_DIR), and so on. This probably also deserves some API support in GLib. > I think we have a lot of disjoint namespaces like wmclass, binary names, > package names, D-Bus names, GSettings schema names, GApplication IDs and > so on. A lot of these are already the same (ie: equal to the D-Bus > name). It would really nice if we could get a strong story for mapping > between these and the rest of them and this could provide that. With systemd I am working on getting a definition of an application into our stack that transcends all its layers. i.e. a cgroup in the kernel, to the .desktop file in the FS, and the windows on screen and so on. It would be fantastic if we had a normalized identifier scheme for this which is accepted on all layers. Lennart -- Lennart Poettering - Red Hat, Inc. _______________________________________________ xdg mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/xdg
