(completion-at-point) brings up menu, but I can't do any with it

[SequentialDesign]

when I call (completion-at-point) interactively, I get a menu, but I can't do anything with it.

the menu is stuck and doesn't do anything. the same thing happened with my real init (the one in the pic is super minimal (just for testing. the only package I have on it is corfu)). I want to call completion-at-point in a function, but I can't get it to work.

I'm using windows 10. I have not tested it on linux yet. have you seen something like this before ?

I tested it on wsl, and the result is the same: completion-at-point gives me a useless menu

[kartikynwa]

I have not tested it on linux yet. have you seen something like this before ?

Happens on Linux too. This does not happen if the popup appears automatically due to corfu-auto but following the steps you mention I encounter the exact same problem.

[minad]

Hi, I've seen such "useless popups" before, but not for a long time. Iirc these were caused by some problems with input/window focus of the child frame, but these issues have been fixed. I cannot reproduce this on my current installation with emacs -Q (Emacs 29.1 X11/Lucid). I enter and evaluate the following in the scratch buffer and Corfu works as expected:

(package-initialize)
(global-corfu-mode)
(def<M-TAB>

I cannot test on Windows but your report is also the first one about problems there. Please try to create a minimal recipe ideally under Linux. Which Emacs version, window manager (kde, gnome, i3, xmonad, ...) and window toolkit do you use (X11+gtk, X11+lucid, X11+notoolkit, X11+pgtk, Wayland+pgtk, ...)? It could be that this issue occurs only under certain special configurations with certain window managers.´

[kartikynwa]

What is bound to here? completion-at-point?

[minad]

The answer is a C-h k away :)

[kartikynwa]

No sorry. I meant if you pressing M-TAB is invoking completion-at-point for you or not. Doing C-h k will tell me what it does for me but not what is going on in your example.

I am trying to create a minimal reproducible example for this btw.

[kartikynwa]

I used this config to reproduce the problem:

(use-package corfu
  :ensure t
  :init
  (global-corfu-mode))

(use-package cape
  :ensure t
  :init
  (add-to-list 'completion-at-point-functions #'cape-dabbrev)
  (add-to-list 'completion-at-point-functions #'cape-file))

(custom-set-variables
 ;; custom-set-variables was added by Custom.
 ;; If you edit it by hand, you could mess it up, so be careful.
 ;; Your init file should contain only one such instance.
 ;; If there is more than one, they won't work right.
 '(package-selected-packages '(corfu)))
(custom-set-faces
 ;; custom-set-faces was added by Custom.
 ;; If you edit it by hand, you could mess it up, so be careful.
 ;; Your init file should contain only one such instance.
 ;; If there is more than one, they won't work right.
 )

Version and system details

• emacs version: GNU Emacs 29.1
• corfu version: 0.38
• cape version: 0.17
• Emacs is using the GTK3 frontend on X, xorg-server version is 21.1.8
• Distro is Void LInux. uname -a output: Linux hostname 6.5.7_1 #1 SMP PREEMPT_DYNAMIC Wed Oct 11 08:20:04 UTC 2023 x86_64 GNU/Linux

This problem only occurs if completion-at-point is called interactively (M-x completion-at-point RET). If I invoke it by pressing M-TAB, everything works perfectly.

[minad]

Thanks! That's a good one! I see the issue now. The clue is the M-x completion-at-point RET which is needed to reproduce this.

[minad]

I pushed a fix in b2b9a23. Please give it a check.

[kartikynwa]

I just tried it briefly in the scratch buffer and it solves the problem. I tried my best but could not cause it. Thanks for fixing it so fast.

[SequentialDesign]

OMG! thank you so much to both of you !! kartikynwa, thank you so much for explaining it and helping to get to the bottom of the issue !

and, minad, thanks for your great packages !

y'all have made my day !!

[kartikynwa]

I hope your day gets even better