alltray barely working in e17

[Martin Koelewijn]

Binary package hint: alltray

Alltray 0.7.2dev (with Vala built from git)
Ubuntu Jaunty 32-bit
E17-svn (svn 41055) (including ecomorph and systray)

If I do alltray -a in gnome-terminal and click on a window (what app I use as victim doesn't seem to matter) it says:

 (alltray:723): Wnck-CRITICAL **: wnck_window_get_pid: assertion `WNCK_IS_WINDOW (window)' failed
 In get_command_line()

Sometimes it shows an icon but when I click that twice the second time it says:

 (alltray:826): Wnck-CRITICAL **: wnck_workspace_activate: assertion `WNCK_IS_WORKSPACE (space)' failed

If I do alltray -p <pid> it gives the same first error without an icon, but if I close the terminal while the process is running the icon shows up in systray and it works. ??? This doesn't work btw if I do alltray -p in a run dialog. Also closing terminal doesn't make alltray -a to work this magically, only alltray -p.

[Martin Koelewijn (vcoolio)]

I'll post this bug under alltray branch, didn't see that this is about alltray package in ubuntu.

[Martin Koelewijn (vcoolio)]

I'll let it go now, don't understand if this is right or not. I'll hear from anyone if I need to file the bug somewhere else.

[Michael B. Trausch (mtrausch)]

Arrrgh, I sent an email to modify this bug and it didn't work.. and I just now noticed. My apologies for not having picked it up sooner. Additional comments forthcoming...

[Michael B. Trausch (mtrausch)]

Can you try using AllTray from the mainline development tree located at lp:alltray and let me know if the problem still persists? There are some bug fixes in trunk that were added just after the 0.7.3dev release, which may prove to have solved the problems described here.

If you still have problems after updating to lp:alltray, can you show the output of the following command when running your program:
 $ ALLTRAY_DEBUG=ALL alltray -D (normal command line arguments here)

That will help me to be able to pinpoint problems.

[Martin Koelewijn (vcoolio)]

Thanks for replying. Because lp:alltray didn't succeed on 'make' I
compiled vala from "bzr branch lp:~vcs-imports/vala/new-trunk". After
that lp:alltray compiled (though with some warnings while the lot of
text was printed, but it completed successfully). I have the same
strange behavior though.

When I run with argument -a:

prompt$ ALLTRAY_DEBUG=ALL alltray -D -a
AllTray 0.7.3dev+, from bzr branch alltray,
  rev-id <email address hidden>
Copyright (c) 2009 Michael B. Trausch <email address hidden>
Licensed under the GNU GPL v3.0 as published by the Free Software
Foundation.

[2009-07-16 13:20:08 CEST] INFO/MISC: Debugging subsystem setup
complete.
[2009-07-16 13:20:08 CEST] INFO/CL: Command line options parsed.
[2009-07-16 13:20:10 CEST] INFO/AH: Looks like we got win 0x00c37225

(alltray:26311): Wnck-CRITICAL **: wnck_window_get_pid: assertion
`WNCK_IS_WINDOW (window)' failed
[2009-07-16 13:20:10 CEST] INFO/AH: Win 0x00c37225 pid = 0
[2009-07-16 13:20:10 CEST] INFO/CL: Evaluating
alltray-internal-fake-process
[2009-07-16 13:20:10 CEST] INFO/CL: Actually processing arg
[2009-07-16 13:20:10 CEST] INFO/CL: Evaluating 0
[2009-07-16 13:20:10 CEST] INFO/CL: Actually processing arg
[2009-07-16 13:20:10 CEST] INFO/PROCESS: run_fake()
[2009-07-16 13:20:10 CEST] INFO/PROCESS: Process 0 now attached.
[2009-07-16 13:20:10 CEST] INFO/APP: Hey hey, creating a new Application
[2009-07-16 13:20:10 CEST] INFO/MISC: delete_pd_window()
[2009-07-16 13:20:11 CEST] INFO/APP: pid 8144 (gnome-terminal) is not
ours.
  Want pid = 0, ppid = 0, or pgid = 26311, have pid = 8144, ppid =
8143, pgid = 8142
[2009-07-16 13:20:11 CEST] INFO/APP: pid 25957 (claws-mail) is not ours.
  Want pid = 0, ppid = 0, or pgid = 26311, have pid = 25957, ppid =
3373, pgid = 25957
[2009-07-16 13:20:11 CEST] INFO/APP: pid 3523 (firefox-3.6) is not ours.
  Want pid = 0, ppid = 0, or pgid = 26311, have pid = 3523, ppid =
3522, pgid = 3522
[2009-07-16 13:20:11 CEST] INFO/APP: pid 9001 (gedit) is not ours.
  Want pid = 0, ppid = 0, or pgid = 26311, have pid = 9001, ppid = 1,
pgid = 8959
[2009-07-16 13:20:11 CEST] INFO/APP: pid 8959 (nautilus) is not ours.
  Want pid = 0, ppid = 0, or pgid = 26311, have pid = 8959, ppid =
3373, pgid = 8959

** (alltray:26311): WARNING **: Application.vala:97: Got PID 0; some
application is buggy!

########################

And when I do with argument -p it works as soon as I close the terminal
where the process is running.

prompt$ ALLTRAY_DEBUG=ALL alltray -D -p 9001
AllTray 0.7.3dev+, from bzr branch alltray,
  rev-id <email address hidden>
Copyright (c) 2009 Michael B. Trausch <email address hidden>
Licensed under the GNU GPL v3.0 as published by the Free Software
Foundation.

[2009-07-16 13:23:19 CEST] INFO/MISC: Debugging subsystem setup
complete. [2009-07-16 13:23:19 CEST] INFO/CL: Command line options
parsed. [2009-07-16 13:23:19 CEST] INFO/CL: Evaluating
alltray-internal-fake-process [2009-07-16 13:23:19 CEST] INFO/CL:
Actually processing arg [2009-07-16 13:23:19 CEST] INFO/CL: Evaluating
9001 [2009-07-16 13:23:19 CEST] INFO/CL: Actually processing arg
[2009-07-16 13:23:19 CEST] INFO/PROCESS: run_fake()
[2009-07...

[Michael B. Trausch (mtrausch)]

With what application does this occur? It looks like the first time you click, you're clicking on an application that has a PID of 0. That will never work, and that is a bug in the application (or the application is _extremely_ old and predates all current desktop standards; still, that means that the application needs to be updated).

Are you clicking on the window proper, or are you clicking on the window decorations? And for the process that you're attaching to in the second attempt, is that the same application? If so, what is it?

[Martin Koelewijn (vcoolio)]

I clicked inside the window, not the titlebar / decoration, like the
popup window tells me. Said behavior occurs with:
Claws Mail 3.7.2
Midori 0.1.7
Gnome-Terminal 2.26.0
e17 settings-manager
It doesn't matter if I use ecomorph session or normal enlightenment.
It works allright on openbox btw. It may very well be an enlightenment
issue, but how to check? I can kill using pids in the terminal
allright, and the pid-flag works with alltray if I close the terminal,
so getting 0 as a pid is not right. If you need more info, let me hear.

Martin

On Mon, 20 Jul 2009 22:37:32 -0000
"Michael B. Trausch" <email address hidden> wrote:

> With what application does this occur? It looks like the first time
> you click, you're clicking on an application that has a PID of 0.
> That will never work, and that is a bug in the application (or the
> application is _extremely_ old and predates all current desktop
> standards; still, that means that the application needs to be
> updated).
>
> Are you clicking on the window proper, or are you clicking on the
> window decorations? And for the process that you're attaching to in
> the second attempt, is that the same application? If so, what is it?
>

[Michael B. Trausch (mtrausch)]

Alright... can you provide me instructions on replicating your setup? Or are you able to provide an account for me that I may SSH into (with a key; you can use the one I have here on LP) where I can use X forwarding and start an Xnest session? This behavior does not seem to exhibit itself in e16 using the 'trayer' system tray utility, though it does appear that e16 doesn't honor the request to hide the application and instead shows it as iconified (I will likely need to file a bug in e16 for that).

[Michael B. Trausch (mtrausch)]

It would appear that e16 and e17 ignore or otherwise do not implement various portions of the the specs that AllTray depends on. I have filed a bug with the Enlightenment people[0] to request that these be implemented. I don't know that e16 will ever get fixed (AllTray's breakage is only minor there), but hopefully they'll fix e17 so that AllTray can properly run on it.

[0]: http://trac.enlightenment.org/e/ticket/373

[Michael B. Trausch (mtrausch)]

This is not a bug in AllTray, so I am marking it invalid here. I will try to follow up with the E developers should they need additional information, of course.

[Martin Koelewijn (vcoolio)]

Ok, thanks for investigating.