|
Bugzilla – Full Text Bug Listing |
| Summary: | dbus-launch does not execute on forwarded X sessions | ||
|---|---|---|---|
| Product: | [openSUSE] SUSE Linux 10.1 | Reporter: | Gary Ekker <gekker> |
| Component: | X.Org | Assignee: | Stefan Dirsch <sndirsch> |
| Status: | RESOLVED WONTFIX | QA Contact: | E-mail List <xorg-maintainer-bugs> |
| Severity: | Enhancement | ||
| Priority: | P5 - None | CC: | eich, sndirsch, werner |
| Version: | unspecified | ||
| Target Milestone: | --- | ||
| Hardware: | Other | ||
| OS: | All | ||
| Whiteboard: | |||
| Found By: | Other | Services Priority: | |
| Business Priority: | Blocker: | --- | |
| Marketing QA Status: | --- | IT Deployment: | --- |
|
Description
Gary Ekker
2005-09-21 20:55:45 UTC
This is really an X issue since all this was pulled down into xinit as a matter of policy. What about giving some more details? Q: Does this mean that we need something like
test -n "$DBUS_SESSION_BUS_ADDRESS"
and hope that we've a running dbus session?
Yes, and if we don't have a running session bus, then we need to launch one. I'm still missing feedback to my comment #2. So dbus-launch is only exec'd with the windowmanager. gekker@hegel:~> env | grep dbus WINDOWMANAGER=/usr/bin/dbus-launch --sh-syntax --exit-with-session /usr/X11R6/bin/gnome DBUS_SESSION_BUS_ADDRESS=unix:abstract=/tmp/dbus-m040xX2zYG,guid=bdd7b243f38b37ca2902fcf8f8a5a200 If one were to "ssh -X hegel.provo.novell.com" and the "env | grep dbus" there is no DBUS_SESSION_BUS_ADDRESS, since we never launched the session bus with dbus-launch. Sorry for the confusion on this bug, what we are asking is, how can we also get the session bus running when on a forwarded X session? Gary, there is on major problem: even if I add a boolean switch to attach on a current dbus session, this IMHO will not help if you're are using ssh to forward the X session. This because as van see above the DBUS_SESSION_BUS_ADDRESS contains a unix domain socket which is only valid on the local host. I think it makes sense to assign this one to Werner. :-) I'm now testing for a valid DBUS_SESSION_BUS_ADDRESS variable but this does not solve the problem with forwarded X sessions. Maybe the ssh command should also export the variable DBUS_SESSION_BUS_ADDRESS ... but nevertheless, a unix domain socket isn't very helpful for a dbus message from the remote to the local dbus-deamon. ------------------------------------------------------------------- Wed Jan 11 13:23:39 CET 2006 - werner@suse.de - Check for an already running dbus session before launching a new session (bug #118291). Thanks! Werner, can we close this now as FIXED? No, this is not fixed. Remaing problem is how to forward a dbus session to a remote host. Currently I do not see how dbus is accessible over `ssh- X'. Timo? Sorry, since I haven't tried this myself I don't know if or how it works. Maybe those pointers are some help: http://lists.freedesktop.org/archives/dbus/2005-February/002141.html http://lists.freedesktop.org/archives/dbus/2005-January/002011.html I take this back as enhancement. Gary, this probably needs to be added to FATE. Otherwise nobody will work on this. JFYI. JFYI, Matthias. This is a bugreport, which is assigned to Egbert/me or with Egbert/me in CC or reported by Egbert/me. Originally dbus has not been desinged for network tranparency like X. Therefore X clients that also require some dbus communication simply cannot be used accross the network. dbux communication has nothing to do with X communicatin therefore it cannot be trivially added. It may be pssible to tunnel it thru X - but this is another area of exploration and not just a bug to fix. Appearantly (according to above links) work is under way to implement network transparent dbus communication however instead of implementing a tcp transport for dbus - following the present trend in X - it should be investigated if it's pssible to forward dbus thru ssh and treat it similarly to X11 forwarding thru the same mechansim. This would of course not address the case where X apps are run accross a network using the tcp transport mechansim - which is however disabled im most case today anyhow. - but this is totally out of the realm of X at present. (In reply to comment #14 from Stefan Dirsch) > I take this back as enhancement. Gary, this probably needs to be added to > FATE. Otherwise nobody will work on this. Since this never happened, it can't be that important. Finally closing as WONTFIX. |