! Please note that this is a snapshot of our old Bugzilla server, which is read only since May 29, 2020. Please go to gitlab.xfce.org for our new server !
Panel startup much slower
Status:
RESOLVED: FIXED
Severity:
critical
Product:
Xfce4-panel

Comments

Description Klaus Kusche 2011-06-15 14:28:30 CEST
In previous 4.8 versions, the panel started extremely fast on login:
By far less than a second, before any other application
(I'm on a fast machine).

In today's git version, the panel takes about 5 seconds to appear at login,
long after any other application started automatically at login.

Moreover, there is a message in .xsession-errors which has not been there
in old versions:
"xfce4-panel: No window manager registered on screen 0. To start the panel without this check, run with --disable-wm-check."
Comment 1 Nick Schermer editbugs 2011-06-15 15:27:20 CEST
And the window manager is running? It should wait for the window manager to appear, if it does the panel is directly started (so we don't switch compositing twice).
Comment 2 Nick Schermer editbugs 2011-06-15 15:29:57 CEST
Do you use xfwm4?
Comment 3 Klaus Kusche 2011-06-15 16:01:35 CEST
Yep, both the old and the new times are with a standard xfwm4 setup:
Things are started by xfce4-session, including xfwm4.
Comment 4 Klaus Kusche 2011-06-15 16:13:26 CEST
Just tried again:
There are more than 3 seconds between the terminal window appearing
(including title bar, hence xfwm is already running) and the panels coming up.

Also, pids suggest that xfwm is started shortly before the panel:
xfwm 12646, then comes Thunar, then the panel (12651).

However, there are *ages* between the panel and the panel plugins:
Panel start time 16:02:07, and the plugins are started at 16:02:13,
beginning at pid 12703.

What is the interval for the window manager check in the panel?
(I mean, if the first check fails as the message shows,
how long does the panel wait before it tries again?)
Comment 5 Nick Schermer editbugs 2011-06-15 17:10:42 CEST
The check is running every 50ms for 5 seconds (before it starts anyway), looking for the _NET_WM_CM_S0 atom (wm in screen 0).
Comment 6 Klaus Kusche 2011-06-15 17:13:41 CEST
So obviously we have the "starts anyway" case.

But what makes the check fail? It worked in 4.8 release.

Is there anything I can test easily?
Comment 7 Nick Schermer editbugs 2011-06-19 13:39:22 CEST
Mmm it depends on compositing enabled or not.
Comment 8 Nick Schermer editbugs 2011-06-19 13:48:25 CEST
Fixed in e3e6be1.
Comment 9 Klaus Kusche 2011-06-19 13:49:28 CEST
I had compositing compiled in but turned off.

Unfortunately, I can't test any more: 
Current git (as provided by Gentoo -9999 ebuilds) doesn't compile for me, 
I'm back at the released versions.

Bug #7727

Reported by:
Klaus Kusche
Reported on: 2011-06-15
Last modified on: 2011-06-19

People

Assignee:
Nick Schermer
CC List:
0 users

Version

Version:
Unspecified

Attachments

Additional information