Why is `after-change-functions' reset?

Why is `after-change-functions' reset?

[Michael Heerdegen]

Hello,

I'm working on a library that adds a function to
`after-change-functions'.

After some time, the thing stops working.  I look at the value of
`after-change-functions' and my added function is missing.

So I thought there would have been some error that caused
`after-change-functions' to be reset.  But I can preclude that - there
is no error, no error message, etc.

I had a hard time to find some recipe for emacs -Q.  Here it comes:

--8<---------------cut here---------------start------------->8---
(progn
  
  (defun foo (&rest _) ())
  
  (add-hook 'after-change-functions 'foo)
  
  (defun complain ()
    (when (not (memq 'foo (default-value 'after-change-functions)))
      (message "Why does this happen?")
      (cancel-timer complain-timer)))
  
  (defvar complain-timer
    (run-with-idle-timer .03 .03 #'complain))

  (require 'cl-lib)
  (completing-read "Hit tab and C-g:  "
		   (cl-loop for i from 1 to 100000
			    collect (format "%d" i))))
--8<---------------cut here---------------end--------------->8---

Evaluate it and hit tab (for completion) and C-g C-g in succession.
When you repeat this a few times, you get "Why does this happen?".

Is this expected behavior?  In my original situation, quitting was not
involved, the above recipe was just the first reproducible program I had
discovered.


Thanks,

Michael.

Re: Why is `after-change-functions' reset?

[Kalle Olavi Niemitalo]

Michael Heerdegen <michael_heerdegen@web.de> writes:

> Evaluate it and hit tab (for completion) and C-g C-g in succession.
> When you repeat this a few times, you get "Why does this happen?".

I debugged this a bit, with my custom Emacs 23.0.51.  I don't
know how much this machinery has changed in later versions.
Emacs set Vquit_flag at:

  handle_interrupt
  tty_read_avail_input
  read_avail_input
  handle_async_input
  input_available_signal
  <signal handler called>
  mark_object
  Fgarbage_collect
  Ffuncall
  run_hook_list_with_args
  signal_after_change
  Fadd_text_properties
  Fput_text_property
  Fdisplay_completion_list
  internal_with_output_to_temp_buffer
  Fminibuffer_completion_help
  do_completion
  Fminibuffer_complete
  Ffuncall
  Fcall_interactively

After Fgarbage_collect returned, Ffuncall called funcall_lambda,
in which the QUIT macro detected the non-nil value of Vquit_flag
and called Fsignal (Qquit, Qnil).  Because the quit was signaled
within signal_after_change, after-change-functions were not
restored.

I don't see any reasonable way to prevent after-change-functions
from being reset like this.

> In my original situation, quitting was not involved, the above
> recipe was just the first reproducible program I had discovered.

Perhaps you could change your Emacs to break when the reset
happens, and always run it in a debugger.  Make
signal_before_change and signal_after_change maintain some kind
of flag or nesting counter, and set conditional breakpoints in
Fthrow and Fsignal.  That will get a false positive if the error
is caught within the hook; but I don't think it happens often.

Re: Why is `after-change-functions' reset?

[Stefan Monnier]

[ Please make it a bug report.  ]

> Evaluate it and hit tab (for completion) and C-g C-g in succession.
> When you repeat this a few times, you get "Why does this happen?".

I couldn't reproduce it.  After the "Why does this happen?", if you do
M-: (default-value 'after-change-functions) RET what do you get?


        Stefan

Re: Why is `after-change-functions' reset?

[Michael Heerdegen]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

> [ Please make it a bug report.  ]

Ok, will do.

> > Evaluate it and hit tab (for completion) and C-g C-g in succession.
> > When you repeat this a few times, you get "Why does this happen?".
>
> I couldn't reproduce it.

It doesn't always happen, but after a few iterations, it'll happen
sooner or later for me.

> After the "Why does this happen?", if you do M-: (default-value
> 'after-change-functions) RET what do you get?

nil.

Michael.

Re: Why is `after-change-functions' reset?

[Michael Heerdegen]

Hi Kalle,

many thanks for debugging.  I've created a bug report, bug#16796, where
I refer to your answer.

Let's continue the discussion in the report's thread.

Michael.