GNU bug report logs - #70186
Variable watcher notification bugs for "local if set" variables. Fix and ert in email to follow. (5 of 9)

Previous Next

To reply to this bug, email your comments to 70186 AT debbugs.gnu.org.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Robert Burks <rburksdev <at> gmail.com>
To: GNU BUGS <bug-gnu-emacs <at> gnu.org>
Subject: Variable watcher notification bugs for "local if set" variables. Fix
 and ert in email to follow. (5 of 9)
Date: Thu, 4 Apr 2024 04:46:07 -0400

[Message part 1 (text/plain, inline)]

(5 of 9)

Bug#00004 (no corresponding unlet notification and improper use of 'where'
            for local_if_set buffer local variables that shadow default.)

** Bug recreations are at the end

I have included a patch for this bug that applies on top of my four previous
submissions in the following email.  This bug occurs because an unaccounted
for
path in do_one_unbind calls set_default_internal.  Current regression
testing
only tests reaching the section of code my patch resides in from a direct
call
using set-default.  I have also included testing that covers all the
examples
provided in the next email.

It was the end of this bug's testing in which I unraveled even more
existing bugs
in which my work did not already solve.  The bug following this required
being
solved in conjunction with this bug.


Bug
Recreation------------------------------------------------------------------

-------------------------------------------
- LIS let and set in the same buffer. (shadows default)
-------------------------------------------
(defvar-local test 100)
test

(defvar results nil)
results

(add-variable-watcher 'test (lambda (&rest args)
                              (push args results)))
nil

(let ((test 99))
  (set 'test 66))
66

results
((test 100 set nil) (test 66 set #<buffer *scratch*>) (test 99 let #<buffer
*scratch*>))
;; there should be an unlet and the others should not have a
'where'!!!!!!!!!
;; Every let should have a corresponding unlet so a watcher function can
relieve
;; it's activity (if so desired) of watching its variable in a particular
buffer
;; and then resume it after the unlet occurs in that particular buffer.
;; Also, in this example the 'let' is acting on the default value as this
is a
;; 'let_shadows_default' case.  The 'let','set', and 'unlet' all act on the
default for
;; all variables that do not have their own value.  It should not have a
'where'.

-------------------------------------------
- LIS let and set in some buffer. (shadows default)
-------------------------------------------
(defvar-local test 5)
test

(defvar results nil)
results

(add-variable-watcher 'test (lambda (&rest args)
                              (push args results)))
nil

(with-temp-buffer
  (let ((test 99))
    (set 'test 66)))
66

results
((test 5 set nil) (test 66 set #<killed buffer>) (test 99 let #<killed
buffer>))

[Message part 2 (text/html, inline)]

Message #8 received at 70186 <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: Robert Burks <rburksdev <at> gmail.com>,
 Stefan Monnier <monnier <at> iro.umontreal.ca>
Cc: 70186 <at> debbugs.gnu.org
Subject: Re: bug#70186: Variable watcher notification bugs for "local if set"
 variables. Fix and ert in email to follow. (5 of 9)
Date: Sat, 06 Apr 2024 10:37:31 +0300

> From: Robert Burks <rburksdev <at> gmail.com>
> Date: Thu, 4 Apr 2024 04:46:07 -0400
> 
> (5 of 9)
> 
> Bug#00004 (no corresponding unlet notification and improper use of 'where'
>             for local_if_set buffer local variables that shadow default.)
> 
> ** Bug recreations are at the end           
> 
> I have included a patch for this bug that applies on top of my four previous
> submissions in the following email.  This bug occurs because an unaccounted for
> path in do_one_unbind calls set_default_internal.  Current regression testing
> only tests reaching the section of code my patch resides in from a direct call
> using set-default.  I have also included testing that covers all the examples
> provided in the next email.
> 
> It was the end of this bug's testing in which I unraveled even more existing bugs
> in which my work did not already solve.  The bug following this required being
> solved in conjunction with this bug.
> 
> Bug Recreation------------------------------------------------------------------
> 
> -------------------------------------------
> - LIS let and set in the same buffer. (shadows default)
> -------------------------------------------
> (defvar-local test 100)
> test
> 
> (defvar results nil)
> results
> 
> (add-variable-watcher 'test (lambda (&rest args)
>                               (push args results)))
> nil
> 
> (let ((test 99))
>   (set 'test 66))
> 66
> 
> results
> ((test 100 set nil) (test 66 set #<buffer *scratch*>) (test 99 let #<buffer *scratch*>))
> ;; there should be an unlet and the others should not have a 'where'!!!!!!!!!
> ;; Every let should have a corresponding unlet so a watcher function can relieve
> ;; it's activity (if so desired) of watching its variable in a particular buffer
> ;; and then resume it after the unlet occurs in that particular buffer.
> ;; Also, in this example the 'let' is acting on the default value as this is a
> ;; 'let_shadows_default' case.  The 'let','set', and 'unlet' all act on the default for
> ;; all variables that do not have their own value.  It should not have a 'where'.
> 
> -------------------------------------------
> - LIS let and set in some buffer. (shadows default)
> -------------------------------------------
> (defvar-local test 5)
> test
> 
> (defvar results nil)
> results
> 
> (add-variable-watcher 'test (lambda (&rest args)
>                               (push args results)))
> nil
> 
> (with-temp-buffer
>   (let ((test 99))
>     (set 'test 66)))
> 66
> 
> results
> ((test 5 set nil) (test 66 set #<killed buffer>) (test 99 let #<killed buffer>))

Stefan, any comments?

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.