Closed paoloschi closed 5 years ago
SMALL UPDATE: I now replaced the terminator application with qterminal and geany with featherpad but unfortunately nothing has changed with regard to the crashes described above...
In both cases I switched from gtk to qt application and found a strange behavior that I'm not sure if it is classifiable as a windowchef issue (to be opened separately):
Below I report my current configuration where I have put right as mouse button for the click_to_focus
setting.
I find that all the windows respond strictly to this setting (only the right button activates them) while both featherpad and qterminal as any other qt5 based application I can bring them focused to the foreground either by pressing the right but unexpectedly also middle and left buttons of the mouse!
waitron wm_config apply_settings true
waitron wm_config enable_borders true
waitron wm_config border_width 5
waitron wm_config internal_border_width 2
waitron wm_config color_focused 0x32AB8B
waitron wm_config color_unfocused 0x404040
waitron wm_config internal_color_focused 0x000000
waitron wm_config internal_color_unfocused 0x6B6B6B
waitron wm_config gap_width all 0
waitron wm_config grid_gap_width 0
waitron wm_config cursor_position center
waitron wm_config pointer_actions move resize_side resize_corner
waitron wm_config pointer_modifier super
waitron wm_config sticky_windows false
waitron wm_config enable_resize_hints false
waitron wm_config enable_sloppy_focus false
waitron wm_config enable_last_window_focusing true
waitron wm_config replay_click_on_focus false
waitron wm_config click_to_focus right
waitron wm_config groups_nr 10
Sorry but new circumstances and evaluations have led me to consider another wm (...) Therefore, in the impossibility of being able to re-discuss the issues that I have raised so far, I think I am doing the right thing in closing this issue as of now. I apologize for the noise.
It is I who needs to apologize here for neglecting your issue. I didn't have the time to investigate it because of school and I don't want to reply without having any concrete answer. I have absolutely no idea why this happens. You are also the only one with this issue, probably. I tried running while :; do waitron window_rev_cycle; done
and it didn't crash at all.
As long as the bug exists, the issue must be open.
You don't have to apologize for the noise, you did well by reporting this error.
Windowchef is a side project of mine that didn't receive much attention from me lately, mostly because I consider it pretty complete, and because of the lack of time. I'm open to contributions, but issues like this can't be reviewed quickly by me, for now. I don't even use it anymore, not because I don't like it but because I didn't want motivation to spend time ricing :)
Here's some information about the WM and OS:
Void Linux x86_64
Kernel: 5.3.7
Nvidia blobs: 430.40
windowchef -v
windowchef v0.4.1-28-g9496a98
Copyright (c) 2016-2017 Tudor Ioan Roman
Released under the ISC License
I also hit the same bug paoloschi described before. I can't seem to find a way to debug it at all. I did tests where I only had 3 rxvt-unicode terminals opened, no panels, no candy. I cycled through windows via 'window_cycle' which seem to work just fine but when I switched to 'window_rev_cycle' instead; I got a crash immediately. I fail to find any logs and errors. For the time being I've disabled a binding for that command from sxhkd.
I'll gladly help but I'm not experienced enough to find the root cause for this by myself. I can run tests whenever needed if you happen to have a hunch about this at some point.
I found a reliable way to reproduce the issue:
If I do the same but with 'waitron window_cycle' instead; crash doesn't happen at all.
Here's the debug log:
loading /home/shn/.config/windowchef/windowchefrc
X Event 33
IPC Command 27 with arguments 0 8 0
X Event 33
IPC Command 27 with arguments 3 2 0
X Event 33
IPC Command 27 with arguments 1 2240053 0
X Event 33
IPC Command 27 with arguments 2 2240053 0
X Event 33
IPC Command 27 with arguments 4 12402495 0
X Event 33
IPC Command 27 with arguments 5 3818848 0
X Event 33
IPC Command 27 with arguments 6 9 0
X Event 33
IPC Command 27 with arguments 7 0 0
X Event 33
IPC Command 27 with arguments 8 4 0
X Event 33
IPC Command 27 with arguments 9 5 0
X Event 33
IPC Command 27 with arguments 10 0 0
X Event 33
IPC Command 27 with arguments 11 0 0
X Event 33
IPC Command 27 with arguments 12 0 0
X Event 33
IPC Command 27 with arguments 13 1 0
X Event 33
IPC Command 27 with arguments 14 1 0
X Event 33
IPC Command 27 with arguments 15 1 0
X Event 33
IPC Command 27 with arguments 16 1 0
X Event 33
IPC Command 27 with arguments 17 2 4
X Event 33
IPC Command 27 with arguments 18 8 0
X Event 33
IPC Command 27 with arguments 19 0 0
X Event 34
X Event 34
X Event 16
X Event 20
new window was born 0x00e00006
X Event 22
X Event 19
grabbed buttons on 0x00e00006
X Event 7
X Event 22
X Event 10
X Event 9
X Event 9
X Event 10
X Event 16
X Event 20
new window was born 0x01000006
X Event 22
X Event 19
grabbed buttons on 0x01000006
X Event 7
X Event 22
X Event 10
X Event 9
X Event 9
X Event 18
grabbed buttons on 0x00e00006
X Event 10
X Event 9
X Event 7
X Event 17
freeing 0x01000006
X Event 10
X Event 9
X Event 0
X Event 33
Your reliable way to reproduce the issue helped me a lot. I quickly found the issue via gdb
and solved it.
Thanks to both of you for reporting!
I am on VoidLinux; the graphical HW is the internal one of an Intel i5 processor with the distro packaged xf86-video-intel Xorg driver. I'm organizing my first windowchef-based configuration and I have very little SW installed in the system at the moment; I'm developing small bash scripts to customize the desktop to my liking (...) Running applications can be SpaceFM Terminator Palemoon Claws-Mail Geany (little more) while at starting session I launch sxhkd ruler clipnotify evtest synergy jgmenu and no panel bar. VoidLinux has windowchef in the repository but since yesterday I have compiled by myself enabling debugging and here I report 5 final parts of the logs happened in less than 12 hours.
Unfortunately I still have not managed to isolate a sequence of actions to make repeatable with ease the crash event. I'm only sure that it never happens with other waitron commands that
window_rev_cycle
, executed indifferently from a script or pressing 'alt+shift+Tab' but I can also runwindow_rev_cycle
a large number of times consecutively and very quickly with success. The crash may occur at the firstwindow_rev_cycle
execution after a series of other waitron commands or after repeated executions of the same command that have previously been successful. From the logs I see that the error is alwaysfatal IO error 11
preceded byX Event 33
(but event 33 occurs also many other times in the entire log without issues!) while out of 5 cases, 4 times it happenedafter 11 requests
... but I am not a programmer: if this information is not enough, I will need precise instructions to provide more useful logs...