Remove icomplete workaround

Remove icomplete workaround

[Aaron Jensen]

[-- Attachment #1: Type: text/plain, Size: 654 bytes --]

Originally added because of: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=38024

The reason this fixed things was not captured and I can not reproduce
the bug with the fix disabled on master.

This workaround has been copied to other libraries without clear
understanding of why it existed in the first place and it caused a
problem: https://github.com/minad/vertico/issues/115#issuecomment-947330581

Would it be possible to remove it? I'm ccing the original reporter and
fixer in case they are able to shed more light on this or test without
the fix to see if they can reproduce it.

I've attached a patch if we are good to remove it.

Thanks,

Aaron

[-- Attachment #2: 0001-lisp-icomplete.el-icomplete-exhibit-Remove-workaroun.patch --]
[-- Type: application/octet-stream, Size: 1113 bytes --]

From 977e40556e9b416e87d9da90b8b5a44068dea378 Mon Sep 17 00:00:00 2001
From: Aaron Jensen <aaronjensen@gmail.com>
Date: Sun, 31 Oct 2021 20:40:43 -0400
Subject: [PATCH] * lisp/icomplete.el (icomplete-exhibit): Remove workaround

---
 lisp/icomplete.el | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/lisp/icomplete.el b/lisp/icomplete.el
index 01033474d3..a61c9d6354 100644
--- a/lisp/icomplete.el
+++ b/lisp/icomplete.el
@@ -716,11 +716,6 @@ icomplete-exhibit
             (delete-region (overlay-start rfn-eshadow-overlay)
                            (overlay-end rfn-eshadow-overlay)))
           (let* ((field-string (icomplete--field-string))
-                 ;; Not sure why, but such requests seem to come
-                 ;; every once in a while.  It's not fully
-                 ;; deterministic but `C-x C-f M-DEL M-DEL ...'
-                 ;; seems to trigger it fairly often!
-                 (while-no-input-ignore-events '(selection-request))
                  (text (while-no-input
                          (icomplete-completions
                           field-string
-- 
2.33.1

Re: Remove icomplete workaround

[Stefan Kangas]

Aaron Jensen <aaronjensen@gmail.com> writes:

> Originally added because of: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=38024

Thanks for the bug report and patch.

If you don't see any follow-ups within the next week or so, please
forward it to bug-gnu-emacs@gnu.org so that we don't lose track of this
issue.  Thanks!

(I can't comment further as I don't know anything about icomplete.el.)

bug#38024: Remove icomplete workaround

[Stefan Monnier via Bug reports for GNU Emacs, the Swiss army knife of text editors]

Aaron Jensen [2021-10-31 20:49:23] wrote:
> Originally added because of: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=38024
>
> The reason this fixed things was not captured and I can not reproduce
> the bug with the fix disabled on master.
>
> This workaround has been copied to other libraries without clear
> understanding of why it existed in the first place and it caused a
> problem: https://github.com/minad/vertico/issues/115#issuecomment-947330581
>
> Would it be possible to remove it? I'm ccing the original reporter and
> fixer in case they are able to shed more light on this or test without
> the fix to see if they can reproduce it.
>
> I've attached a patch if we are good to remove it.

Thanks, I just pushed your patch to `master`.
We'll see if the problem comes back, and if so, we'll have to try and
fix it better.


        Stefan

bug#38024: Remove icomplete workaround

[Aaron Jensen]

On Sun, Oct 31, 2021 at 10:38 PM Stefan Monnier
<monnier@iro.umontreal.ca> wrote:
>
> Aaron Jensen [2021-10-31 20:49:23] wrote:
> > Originally added because of: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=38024
> >
> > The reason this fixed things was not captured and I can not reproduce
> > the bug with the fix disabled on master.
> >
> > This workaround has been copied to other libraries without clear
> > understanding of why it existed in the first place and it caused a
> > problem: https://github.com/minad/vertico/issues/115#issuecomment-947330581
> >
> > Would it be possible to remove it? I'm ccing the original reporter and
> > fixer in case they are able to shed more light on this or test without
> > the fix to see if they can reproduce it.
> >
> > I've attached a patch if we are good to remove it.
>
> Thanks, I just pushed your patch to `master`.
> We'll see if the problem comes back, and if so, we'll have to try and
> fix it better.

Great, thank you!

bug#38024: Remove icomplete workaround

[Kévin Le Gouguec]

Hey Aaron,

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

> Aaron Jensen [2021-10-31 20:49:23] wrote:
>>
>> Would it be possible to remove it? I'm ccing the original reporter and
>> fixer in case they are able to shed more light on this or test without
>> the fix to see if they can reproduce it.
>>
>> I've attached a patch if we are good to remove it.
>
> Thanks, I just pushed your patch to `master`.
> We'll see if the problem comes back, and if so, we'll have to try and
> fix it better.

Congrats on your investigation work; I wish I could shed more light on
what this workaround was meant to accomplish.  I've tried to reproduce
my original issue, but so far I was not successful.

If my initial report can be trusted, the bug happened (1) back when the
development branch was 27.0.50 (commit 61fb521481) (2) on an old
battered laptop that I can no longer power on without disturbing loud
fan demons.

26.3 no longer compiles on my end[1], so all I could do was try to
reproduce on master, where the workaround has been reverted, on other
computers I have lying around.  The symptoms have not resurfaced yet,
so… I'll go with a timid, yet hopeful "yay" 🙂


[1] With gcc 11.2.1:
sysdep.c:1795:22: error: variably modified ‘sigsegv_stack’ at file scope
 1795 | static unsigned char sigsegv_stack[SIGSTKSZ];
      |                      ^~~~~~~~~~~~~

bug#38024: Remove icomplete workaround

[Aaron Jensen]

On Mon, Nov 1, 2021 at 5:28 AM Kévin Le Gouguec
<kevin.legouguec@gmail.com> wrote:
>
> Hey Aaron,
>
> Stefan Monnier <monnier@iro.umontreal.ca> writes:
>
> > Aaron Jensen [2021-10-31 20:49:23] wrote:
> >>
> >> Would it be possible to remove it? I'm ccing the original reporter and
> >> fixer in case they are able to shed more light on this or test without
> >> the fix to see if they can reproduce it.
> >>
> >> I've attached a patch if we are good to remove it.
> >
> > Thanks, I just pushed your patch to `master`.
> > We'll see if the problem comes back, and if so, we'll have to try and
> > fix it better.
>
> Congrats on your investigation work; I wish I could shed more light on
> what this workaround was meant to accomplish.  I've tried to reproduce
> my original issue, but so far I was not successful.
>
> If my initial report can be trusted, the bug happened (1) back when the
> development branch was 27.0.50 (commit 61fb521481) (2) on an old
> battered laptop that I can no longer power on without disturbing loud
> fan demons.
>
> 26.3 no longer compiles on my end[1], so all I could do was try to
> reproduce on master, where the workaround has been reverted, on other
> computers I have lying around.  The symptoms have not resurfaced yet,
> so… I'll go with a timid, yet hopeful "yay" 🙂

Thank you for confirming. Hopefully that's the last of it!

Best,

Aaron