Can't switch away from some workspaces with single X window #688
Comments
|
To my untrained ears, the description sounds similar to #617 (comment). If you look at the issue, I have linked to a screen-recording of the issue.
|
|
You may also consider sharing See |
|
@dto Do you have a multi-monitor setup or is |
|
I can confirm this bug. I have two workspaces on differen screens. Buffer firefox<2> is on workspace 0 and scratch is on workspace 1. When I try to switch from workspace 0 to 1 with function `exwm-workspace-switch' focus jump back to workspace 0. Here is attached xelb debug log, where I try to change workspace four times: Answers to second comment: Note: This happens only if frame with firefox has only one window. If I split frame into two windows (with at least one buffer which is not exwm) this doesn't happen. |
|
The problem is with the logs like: I'm not sure why (advice-add #'exwm-input--on-buffer-list-update :before
(lambda (&rest r)
(exwm--log "CALL STACK: %s" (cddr (reverse (xcb-debug:-call-stack)))))) |
|
Here is the log with call stack: xelb.log |
|
I figured out something. If I have in configuration this snippet, which show me whenever JBoss is running, the problem is present when I try to change workspace with mouse click too. If I put this out, the problem is only present when I call `exwm-workspace-switch' command interactively. |
|
@mobid From the logs it's clearly |
|
Thank you for solution. It works fine now. I was also modify |
|
For me, the issue appears to be eyebrowse. Disabling eyebrowse fixes the issue. log[02:41:19] : CALL STACK: ((t exwm-input--on-buffer-list-update) (t run-hooks buffer-list-update-hook) (t get-buffer-create *temp*) (t generate-new-buffer *temp*) (t format-spec %s ((115 . 1) (116 . ))) (t eyebrowse-format-slot (1 (((min-height . 4) (min-width . 10) (min-height-ignore . 2) (min-width-ignore . 4) (min-height-safe . 1) (min-width-safe . 2) (min-pixel-height . 104) (min-pixel-width . 120) (min-pixel-height-ignore . 52) (min-pixel-width-ignore . 40) (min-pixel-height-safe . 26) (min-pixel-width-safe . 24)) leaf (pixel-width . 3840) (pixel-height . 3014) (total-width . 320) (total-height . 115) (normal-height . 1.0) (normal-width . 1.0) (buffer *spacemacs* (selected . t) (hscroll . 0) (fringes 8 8 nil nil) (margins nil) (scroll-bars nil 0 t nil 0 t nil) (vscroll . 0) (dedicated) (point . 29) (start . 1))) )) (t #[257 \211@\211\300U\203 \0\301\202 \0\302\303 \304�\305\306\307\310\311\312� !\313"\314\315\316&�#\210\211\262�\317\320�!\321�\322��\323� \324\325\326��\327� &�\207 [1 eyebrowse-mode-line-active eyebrowse-mode-line-inactive make-sparse-keymap define-key [mode-line mouse-1] make-byte-code 257 \301\300!\207 vconcat vector [eyebrowse-switch-to-window-config] 3 |
|
Something strange: I fixed a separate issue where my modeline, doom, wasn't being updated by adding: (with-eval-after-load 'doom-modeline
(add-hook 'exwm-workspace-switch-hook #'doom-modeline-set-selected-window))This appears to have fixed the problem. So I'm not sure where the problem is coming from. |
|
I have no clue what side effect |
|
@Stebalien, the reason is that eyebrowse invokes @Stebalien Please, test #737 with eyebrowse and report back. |
|
#737 looks overall good to me. Let's wait for feed backs from @Stebalien and @dto before it's merged. |
|
Sorry for the delay. I've tested #737 and, while it works, it doesn't fix the bug for me. |
|
Is there any workaround to this so far? |
|
@Stebalien Could you regenerate debug logs with #737 patched? @divansantana Which problem did you refer to? What other three users reported may not be the same thing after all. You may check if #737 works for you of course. |
|
@ch11ng This only happens with eyebrowse-mode in this scenario so far, no I'll test your patch as soon as I can. |
|
I've tried using this PR #737 with these two commits last in it. I still have the same issue described above. I've tried to enable debugging and attached it:
Let me know if you need anything else: |
|
My eyebrowse config is pretty standard: (use-package eyebrowse
:ensure t
:pin melpa
:config
(setq-default eyebrowse-mode-line-style 'smart)
(exwm-input-set-key (kbd "s-1") #'eyebrowse-switch-to-window-config-1)
(exwm-input-set-key (kbd "s-2") #'eyebrowse-switch-to-window-config-2)
(exwm-input-set-key (kbd "s-3") #'eyebrowse-switch-to-window-config-3)
(exwm-input-set-key (kbd "s-4") #'eyebrowse-switch-to-window-config-4)
(eyebrowse-mode t)) |
|
Just as @medranocalvo have mentioned, the problem with eyebrowse is with the temp buffer generated during workspace switch. You may apply the following patch as a workaround, which unfortunately conflicts with #737 so for those already applied #737 you'll have to revert it first. We may end up with this as the solution if we can't come up with a better one. diff --git a/exwm-input.el b/exwm-input.el
index decfc8128..9074bd16e 100644
--- a/exwm-input.el
+++ b/exwm-input.el
@@ -1133,6 +1133,19 @@ One use is to access the keymap bound to KEYS (as prefix keys) in char-mode."
(unless (current-message)
(exwm-input--on-minibuffer-exit)))
+(defun exwm-input--ignore-temp-buffer (orig-fun &rest params)
+ "Set `exwm-input--skip-buffer-list-update' to ignore temp buffers."
+ (exwm--log)
+ (let ((exwm-input--skip-buffer-list-update
+ (equal " *temp*" (cond
+ ((stringp (car params))
+ (car params))
+ ((bufferp (car params))
+ (buffer-name (car params)))
+ (t
+ nil)))))
+ (apply orig-fun params)))
+
(defun exwm-input--init ()
"Initialize the keyboard module."
(exwm--log)
@@ -1178,6 +1191,8 @@ One use is to access the keymap bound to KEYS (as prefix keys) in char-mode."
(when mouse-autoselect-window
(xcb:+event exwm--connection 'xcb:EnterNotify
#'exwm-input--on-EnterNotify))
+ (advice-add 'generate-new-buffer :around #'exwm-input--ignore-temp-buffer)
+ (advice-add 'kill-buffer :around #'exwm-input--ignore-temp-buffer)
;; Control `exwm-input--during-command'
(add-hook 'pre-command-hook #'exwm-input--on-pre-command)
(add-hook 'post-command-hook #'exwm-input--on-post-command)
@@ -1199,6 +1214,8 @@ One use is to access the keymap bound to KEYS (as prefix keys) in char-mode."
"Exit the input module."
(exwm--log)
(exwm-input--unset-simulation-keys)
+ (advice-remove 'generate-new-buffer #'exwm-input--ignore-temp-buffer)
+ (advice-remove 'kill-buffer #'exwm-input--ignore-temp-buffer
(remove-hook 'pre-command-hook #'exwm-input--on-pre-command)
(remove-hook 'post-command-hook #'exwm-input--on-post-command)
(remove-hook 'minibuffer-setup-hook #'exwm-input--on-minibuffer-setup) |
|
Seem to have it isolated to:
Anything I can provide? |
|
I can confirm neither the above diff nor #737 resolve this for me. |
I switched to Emacs 27.1 (from Git head) and have a problem. When I try to switch to another workspace via Super-1/2/3 etc, when using whichever is the most recently started X window, the workspace switch happens for a split second and then I am pulled back to the original workspace with the X window. To be able to switch workspaces, I have to hide the X window first.
The text was updated successfully, but these errors were encountered: