* Ignore spurious focus events for ‘after-focus-change-function’
@ 2021-01-15 20:26 Andrea Greselin
2021-01-16 7:02 ` Eli Zaretskii
2021-01-16 9:56 ` Andrea Greselin
0 siblings, 2 replies; 4+ messages in thread
From: Andrea Greselin @ 2021-01-15 20:26 UTC (permalink / raw)
To: help-gnu-emacs
Hello everyone, I'm on Emacs 27.1, Fedora 33, GNOME 3.38.2.
In the latest ‘NEWS’ it is reported that
> The hooks ‘focus-in-hook’ and ‘focus-out-hook’ are now obsolete.
> Instead, attach to ‘after-focus-change-function’ using
> ‘add-function’ and inspect the focus state of each frame using
> ‘frame-focus-state’.
The documentation for ‘after-focus-change-function’ says
> Depending on window system, focus events may also be delivered
> repeatedly and with different focus states before settling to the
> expected values. Code relying on focus notifications should
> "debounce" any user-visible updates arising from focus changes,
> perhaps by deferring work until redisplay.
Indeed, after evaluating this code,
(defun focus-test ()
(message "ffs: %s" (frame-focus-state)))
(add-function :after after-focus-change-function #'focus-test)
every time I switch buffer by selecting one with the mouse my message
buffer shows
ffs: nil
ffs: t
ffs: nil
ffs: t
That is, Emacs sees two back and forth focus changes where I did none,
and most importantly it has hallucinations of focus-out events which
defeat the usefulness of the
‘after-focus-change-function’+‘frame-focus-state’ method for
triggering functions based on focus events.
My question is, how do I "'debounce' any user-visible updates arising
from focus changes"? I made a couple of attempts, with
(defun focus-test ()
(sit-for 0)
(message "ffs: %s" (frame-focus-state)))
and
(defun payload (_window)
(message "ffs: %s" (frame-focus-state))
(remove-hook 'pre-redisplay-functions #'payload))
(defun focus-test ()
(add-hook 'pre-redisplay-functions #'payload))
(add-function :after after-focus-change-function #'focus-test)
but it didn't work.
Andrea
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Ignore spurious focus events for ‘after-focus-change-function’
2021-01-15 20:26 Ignore spurious focus events for ‘after-focus-change-function’ Andrea Greselin
@ 2021-01-16 7:02 ` Eli Zaretskii
2021-01-16 9:56 ` Andrea Greselin
1 sibling, 0 replies; 4+ messages in thread
From: Eli Zaretskii @ 2021-01-16 7:02 UTC (permalink / raw)
To: help-gnu-emacs
> From: Andrea Greselin <greselin.andrea@gmail.com>
> Date: Fri, 15 Jan 2021 21:26:32 +0100
>
> My question is, how do I "'debounce' any user-visible updates arising
> from focus changes"? I made a couple of attempts, with
>
> (defun focus-test ()
> (sit-for 0)
> (message "ffs: %s" (frame-focus-state)))
>
> and
>
> (defun payload (_window)
> (message "ffs: %s" (frame-focus-state))
> (remove-hook 'pre-redisplay-functions #'payload))
> (defun focus-test ()
> (add-hook 'pre-redisplay-functions #'payload))
> (add-function :after after-focus-change-function #'focus-test)
>
> but it didn't work.
What do you want to accomplish, and how did it not work?
In any case, I very much doubt that "deferring work until redisplay"
meant that you should _trigger_ redisplay by calling sit-for. It
probably meant running your code from some hook called by redisplay,
or something similar.
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Ignore spurious focus events for ‘after-focus-change-function’
2021-01-15 20:26 Ignore spurious focus events for ‘after-focus-change-function’ Andrea Greselin
2021-01-16 7:02 ` Eli Zaretskii
@ 2021-01-16 9:56 ` Andrea Greselin
2021-01-18 19:09 ` Andrea Greselin
1 sibling, 1 reply; 4+ messages in thread
From: Andrea Greselin @ 2021-01-16 9:56 UTC (permalink / raw)
To: Help Gnu Emacs mailing list
Hi Eli, I need it for triggering auto-save on focus-out and functions
that dim some colours on focus-out and restore them on focus-in.
Auto-save works when it should, and also when it shouldn't. Changing
the colours on focus out works well but there's some flickering when
they're restored on focus in, which I think may be due to the excess
focus events triggering it repeatedly.
Also, I have an issue with the tab-bar when the Solarized theme is
enabled (the GUI keys -- new tab, close tab, the tab itself -- don't
work when I click on them) and I'm trying to debug it. It must be
something in my configuration because there are no problems with
Solarized in ‘emacs -Q’. I wanted to ignore the spurious focus events
as part of debugging this issue.
> In any case, I very much doubt that "deferring work until redisplay"
> meant that you should _trigger_ redisplay by calling sit-for.
I had tried with ‘(sit-for 0 t)’ first, wich *maybe* was bit cleverer
but yes, it was naive. Now I have something that seems to be working,
from a suggestion on Reddit [1]
(defvar focus-out-auto-save-timer nil)
(defun focus-test () (message "ffs: %s" (frame-focus-state)) (setq
focus-out-auto-save-timer nil))
(defun focus-test-wrapper () (setq focus-out-auto-save-timer (unless
(timerp focus-out-auto-save-timer) (run-at-time "0.05 sec" nil
#'focus-test))))
(add-function :after after-focus-change-function
#'focus-test-wrapper)
It's not super-robust (works with delay >= 0.05 s, doesn't work if
it's <= 0.01 s) but it's done its job for the couple of times I tested
it. In particular, it does not see focus-out events unless one
actually happens. I still hope there is a cleaner way, I don't like
this one very much.
1
https://www.reddit.com/r/emacs/comments/kxsgtn/ignore_spurious_focus_events_for/
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Ignore spurious focus events for ‘after-focus-change-function’
2021-01-16 9:56 ` Andrea Greselin
@ 2021-01-18 19:09 ` Andrea Greselin
0 siblings, 0 replies; 4+ messages in thread
From: Andrea Greselin @ 2021-01-18 19:09 UTC (permalink / raw)
To: Help Gnu Emacs mailing list
I've settled with these home-made remakes of the obsoleted hooks:
;; Set up hooks to be run on focus in and focus out
;; Mutter sends spurious focus events, including sending focus out
;; events when Emacs never really was unfocused.
;; The timer filters out these false focus out events, while
;; checking the ‘last-focus-state’ allows ignoring repeated events
;; of the same kind.
;; REVIEW: false focus out events are still received sometimes when
;; using the menu bar.
;; Adapted from a suggestion by DasEwigeLicht on
;;
https://www.reddit.com/r/emacs/comments/kxsgtn/ignore_spurious_focus_events_for/
(defvar focus-events-timer nil)
(defvar last-focus-state (frame-focus-state))
(defvar my-focus-in-hook nil
"Normal hook run when all frames lose input focus.")
(defvar my-focus-out-hook nil
"Normal hook run when a frame gains focus.")
(defun change-of-focus-functions ()
(unless (equal last-focus-state (frame-focus-state))
(if (frame-focus-state)
(run-hooks 'my-focus-in-hook)
(run-hooks 'my-focus-out-hook)))
(setq focus-events-timer nil)
(setq last-focus-state (frame-focus-state)))
(defun run-change-of-focus-functions-with-timer ()
(setq focus-events-timer
(unless (timerp focus-events-timer)
(run-at-time "0.05 sec" nil ; Found by trial and error.
#'change-of-focus-functions))))
(add-function :after after-focus-change-function
#'run-change-of-focus-functions-with-timer)
;; Test
(defun focus-in-test ()
(message "focus-in"))
(defun focus-out-test ()
(message "focus-out"))
(add-hook 'my-focus-in-hook #'focus-in-test)
(add-hook 'my-focus-out-hook #'focus-out-test)
;; Actual use
(add-hook 'my-focus-out-hook #'do-auto-save)
Please reply with suggestions if there's something wrong or
less-than-optimal.
Andrea
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2021-01-18 19:09 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2021-01-15 20:26 Ignore spurious focus events for ‘after-focus-change-function’ Andrea Greselin
2021-01-16 7:02 ` Eli Zaretskii
2021-01-16 9:56 ` Andrea Greselin
2021-01-18 19:09 ` Andrea Greselin
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).