* bug#46912: 28.0.50; Byte compiled defvar with lexical binding throws (void-variable foo)
@ 2021-03-04 0:14 Vitalie Spinu
2021-03-04 18:03 ` Joseph Mingrone
0 siblings, 1 reply; 13+ messages in thread
From: Vitalie Spinu @ 2021-03-04 0:14 UTC (permalink / raw)
To: 46912
Hello,
Compiling the following file:
;;; -*- lexical-binding: t -*-
(let ((foo t))
(defvar tmp-bar (if foo 0 1 -1)))
gives:
tmp.el:4:23: Warning: reference to free variable ‘foo’
When loading the .elc file the error
Debugger entered--Lisp error: (void-variable foo)
byte-code("\301\10\203\11\0\302\202\n\0\303!\207" [foo (lambda
(def-tmp-var) (defvar tmp-bar def-tmp-var)) 0 -1] 2)
load(".../tmp.elc" nil nil t)
No problems with emacs-27 branch.
Vitalie
In GNU Emacs 28.0.50 (build 1, x86_64-pc-linux-gnu, GTK+ Version
3.22.30, cairo version 1.15.10)
of 2021-03-03 built on NB277
Repository revision: a350ae058caedcb7be7d332564817954e3624e60
Repository branch: HEAD
Windowing system distributor 'The X.Org Foundation', version 11.0.11906000
System Description: Ubuntu 18.04.5 LTS
^ permalink raw reply [flat|nested] 13+ messages in thread
* bug#46912: 28.0.50; Byte compiled defvar with lexical binding throws (void-variable foo)
2021-03-04 0:14 bug#46912: 28.0.50; Byte compiled defvar with lexical binding throws (void-variable foo) Vitalie Spinu
@ 2021-03-04 18:03 ` Joseph Mingrone
2021-03-04 20:09 ` Lars Ingebrigtsen
0 siblings, 1 reply; 13+ messages in thread
From: Joseph Mingrone @ 2021-03-04 18:03 UTC (permalink / raw)
To: 46912; +Cc: Vitalie Spinu, emacs-devel
On Thu, 2021-03-04 at 01:14, Vitalie Spinu <spinuvit@gmail.com> wrote:
> Hello,
> Compiling the following file:
> ;;; -*- lexical-binding: t -*-
> (let ((foo t))
> (defvar tmp-bar (if foo 0 1 -1)))
> gives:
> tmp.el:4:23: Warning: reference to free variable ‘foo’
> When loading the .elc file the error
> Debugger entered--Lisp error: (void-variable foo)
> byte-code("\301\10\203\11\0\302\202\n\0\303!\207" [foo (lambda
> (def-tmp-var) (defvar tmp-bar def-tmp-var)) 0 -1] 2)
> load(".../tmp.elc" nil nil t)
> No problems with emacs-27 branch.
> Vitalie
> In GNU Emacs 28.0.50 (build 1, x86_64-pc-linux-gnu, GTK+ Version
> 3.22.30, cairo version 1.15.10)
> of 2021-03-03 built on NB277
> Repository revision: a350ae058caedcb7be7d332564817954e3624e60
> Repository branch: HEAD
> Windowing system distributor 'The X.Org Foundation', version 11.0.11906000
> System Description: Ubuntu 18.04.5 LTS
Hi,
With
;;; -*- lexical-binding: t -*-
(let ((foo t))
(defvar tmp-bar foo))
the same warning is reported when compiling, but when loading the .elc
file, the debugger isn't started. Just
load: Symbol’s value as variable is void: foo
is reported.
With
;;; -*- lexical-binding: t -*-
(defvar foo
(let ((bar t))
(if bar 1 0)))
The .elc loads without errors.
Joe
^ permalink raw reply [flat|nested] 13+ messages in thread
* bug#46912: 28.0.50; Byte compiled defvar with lexical binding throws (void-variable foo)
2021-03-04 18:03 ` Joseph Mingrone
@ 2021-03-04 20:09 ` Lars Ingebrigtsen
2021-03-04 20:17 ` Joseph Mingrone
2021-03-04 20:21 ` Pip Cet
0 siblings, 2 replies; 13+ messages in thread
From: Lars Ingebrigtsen @ 2021-03-04 20:09 UTC (permalink / raw)
To: Joseph Mingrone; +Cc: 46912, Pip Cet, Vitalie Spinu
Joseph Mingrone <jrm@ftfl.ca> writes:
> With
>
> ;;; -*- lexical-binding: t -*-
>
> (let ((foo t))
> (defvar tmp-bar foo))
>
> the same warning is reported when compiling, but when loading the .elc
> file, the debugger isn't started. Just
>
> load: Symbol’s value as variable is void: foo
>
> is reported.
I can reproduce this problem here, too (and Emacs 27 doesn't have this
problem).
I have not tried to bisect this, but since Pip pushed a change related
to this a couple days ago, I've added Pip to the CCs.
--
(domestic pets only, the antidote for overdose, milk.)
bloggy blog: http://lars.ingebrigtsen.no
^ permalink raw reply [flat|nested] 13+ messages in thread
* bug#46912: 28.0.50; Byte compiled defvar with lexical binding throws (void-variable foo)
2021-03-04 20:09 ` Lars Ingebrigtsen
@ 2021-03-04 20:17 ` Joseph Mingrone
2021-03-04 20:48 ` Lars Ingebrigtsen
2021-03-04 20:21 ` Pip Cet
1 sibling, 1 reply; 13+ messages in thread
From: Joseph Mingrone @ 2021-03-04 20:17 UTC (permalink / raw)
To: Lars Ingebrigtsen; +Cc: 46912, Pip Cet, Vitalie Spinu
On Thu, 2021-03-04 at 21:09, Lars Ingebrigtsen <larsi@gnus.org> wrote:
> Joseph Mingrone <jrm@ftfl.ca> writes:
>> With
>> ;;; -*- lexical-binding: t -*-
>> (let ((foo t))
>> (defvar tmp-bar foo))
>> the same warning is reported when compiling, but when loading the .elc
>> file, the debugger isn't started. Just
>> load: Symbol’s value as variable is void: foo
>> is reported.
> I can reproduce this problem here, too (and Emacs 27 doesn't have this
> problem).
> I have not tried to bisect this, but since Pip pushed a change related
> to this a couple days ago, I've added Pip to the CCs.
I think the problem was introduced before 9b4a2dd (2021-02-14). That
was the last snapshot that I was running and I'm pretty sure I saw the
problem then. If I can find time tonight, I'll try to confirm that.
Joe
^ permalink raw reply [flat|nested] 13+ messages in thread
* bug#46912: 28.0.50; Byte compiled defvar with lexical binding throws (void-variable foo)
2021-03-04 20:09 ` Lars Ingebrigtsen
2021-03-04 20:17 ` Joseph Mingrone
@ 2021-03-04 20:21 ` Pip Cet
1 sibling, 0 replies; 13+ messages in thread
From: Pip Cet @ 2021-03-04 20:21 UTC (permalink / raw)
To: Lars Ingebrigtsen; +Cc: Joseph Mingrone, 46912, Pip Cet, Vitalie Spinu
On Thu, Mar 4, 2021 at 8:10 PM Lars Ingebrigtsen <larsi@gnus.org> wrote:
> Joseph Mingrone <jrm@ftfl.ca> writes:
> > With
> >
> > ;;; -*- lexical-binding: t -*-
> >
> > (let ((foo t))
> > (defvar tmp-bar foo))
> >
> > the same warning is reported when compiling, but when loading the .elc
> > file, the debugger isn't started. Just
> >
> > load: Symbol’s value as variable is void: foo
> >
> > is reported.
>
> I can reproduce this problem here, too (and Emacs 27 doesn't have this
> problem).
I can reproduce it, three, but I can also reproduce it with the change
I think you're referring to reverted.
> I have not tried to bisect this, but since Pip pushed a change related
> to this a couple days ago, I've added Pip to the CCs.
Thanks! I must confess I don't know whether I need to do anything to
make the savannah.gnu.org email address work, but then I need to get a
better email address anyway...
Pip (at mumble, mumble, yes, I do realize that's a non-free email service)
^ permalink raw reply [flat|nested] 13+ messages in thread
* bug#46912: 28.0.50; Byte compiled defvar with lexical binding throws (void-variable foo)
2021-03-04 20:17 ` Joseph Mingrone
@ 2021-03-04 20:48 ` Lars Ingebrigtsen
2021-03-04 21:08 ` Pip Cet
0 siblings, 1 reply; 13+ messages in thread
From: Lars Ingebrigtsen @ 2021-03-04 20:48 UTC (permalink / raw)
To: Joseph Mingrone; +Cc: 46912, Vitalie Spinu, Pip Cet
Joseph Mingrone <jrm@ftfl.ca> writes:
> I think the problem was introduced before 9b4a2dd (2021-02-14). That
> was the last snapshot that I was running and I'm pretty sure I saw the
> problem then. If I can find time tonight, I'll try to confirm that.
Great; bisecting this would be very helpful.
Pip Cet <pipcet@gmail.com> writes:
> I can reproduce it, three, but I can also reproduce it with the change
> I think you're referring to reverted.
Sorry about that; it was a stab in the dark using the "the one who
touched it last" principle. :-/
--
(domestic pets only, the antidote for overdose, milk.)
bloggy blog: http://lars.ingebrigtsen.no
^ permalink raw reply [flat|nested] 13+ messages in thread
* bug#46912: 28.0.50; Byte compiled defvar with lexical binding throws (void-variable foo)
2021-03-04 20:48 ` Lars Ingebrigtsen
@ 2021-03-04 21:08 ` Pip Cet
2021-03-04 23:15 ` Joseph Mingrone
2021-03-05 9:10 ` Mattias Engdegård
0 siblings, 2 replies; 13+ messages in thread
From: Pip Cet @ 2021-03-04 21:08 UTC (permalink / raw)
To: Lars Ingebrigtsen, Mattias; +Cc: Joseph Mingrone, 46912, Vitalie Spinu
[-- Attachment #1: Type: text/plain, Size: 968 bytes --]
On Thu, Mar 4, 2021 at 8:48 PM Lars Ingebrigtsen <larsi@gnus.org> wrote:
> Joseph Mingrone <jrm@ftfl.ca> writes:
> > I think the problem was introduced before 9b4a2dd (2021-02-14). That
> > was the last snapshot that I was running and I'm pretty sure I saw the
> > problem then. If I can find time tonight, I'll try to confirm that.
>
> Great; bisecting this would be very helpful.
Absolutely.
> Pip Cet <pipcet@gmail.com> writes:
>
> > I can reproduce it, three, but I can also reproduce it with the change
> > I think you're referring to reverted.
>
> Sorry about that; it was a stab in the dark using the "the one who
> touched it last" principle. :-/
Not a problem at all. In fact, I've taken a look and it seems
byte-optimize-form-code-walker simply ignores the value form in a
defvar now..
So I suspect it's commit 83983b6b7a115474572973b62eb5e42251713e63, I'm
cc'ing Mattias, and I suspect this patch might help (but I've only
done one quick test).
Pip
[-- Attachment #2: 0001-Don-t-ignore-lexically-bound-variables-in-a-defvar-b.patch --]
[-- Type: text/x-patch, Size: 974 bytes --]
From c39ecf1c053f551a4b384d2ad2994fe0e58d67b0 Mon Sep 17 00:00:00 2001
From: Pip Cet <pipcet@gmail.com>
Date: Thu, 4 Mar 2021 21:06:13 +0000
Subject: [PATCH] Don't ignore lexically-bound variables in a defvar
(bug#46912)
* lisp/emacs-lisp/byte-opt.el (byte-optimize-form-code-walker): Walk
the value form of a defvar.
---
lisp/emacs-lisp/byte-opt.el | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/lisp/emacs-lisp/byte-opt.el b/lisp/emacs-lisp/byte-opt.el
index b51ba801552d6..ba2165a5d59ce 100644
--- a/lisp/emacs-lisp/byte-opt.el
+++ b/lisp/emacs-lisp/byte-opt.el
@@ -607,7 +607,9 @@ byte-optimize-form-code-walker
(setq args (cddr args)))
(cons fn (nreverse var-expr-list))))
- (`(defvar ,(and (pred symbolp) name) . ,_)
+ (`(defvar ,(and (pred symbolp) name) . ,rest)
+ (when rest
+ (byte-optimize-form (car rest) nil))
(push name byte-optimize--dynamic-vars)
form)
--
2.30.1
^ permalink raw reply related [flat|nested] 13+ messages in thread
* bug#46912: 28.0.50; Byte compiled defvar with lexical binding throws (void-variable foo)
2021-03-04 21:08 ` Pip Cet
@ 2021-03-04 23:15 ` Joseph Mingrone
2021-03-05 2:52 ` Pip Cet
2021-03-05 9:10 ` Mattias Engdegård
1 sibling, 1 reply; 13+ messages in thread
From: Joseph Mingrone @ 2021-03-04 23:15 UTC (permalink / raw)
To: Pip Cet; +Cc: Mattias, Lars Ingebrigtsen, Vitalie Spinu, 46912
On Thu, 2021-03-04 at 21:08, Pip Cet <pipcet@gmail.com> wrote:
> On Thu, Mar 4, 2021 at 8:48 PM Lars Ingebrigtsen <larsi@gnus.org> wrote:
>> Joseph Mingrone <jrm@ftfl.ca> writes:
>> > I think the problem was introduced before 9b4a2dd (2021-02-14). That
>> > was the last snapshot that I was running and I'm pretty sure I saw the
>> > problem then. If I can find time tonight, I'll try to confirm that.
>> Great; bisecting this would be very helpful.
> Absolutely.
>> Pip Cet <pipcet@gmail.com> writes:
>> > I can reproduce it, three, but I can also reproduce it with the change
>> > I think you're referring to reverted.
>> Sorry about that; it was a stab in the dark using the "the one who
>> touched it last" principle. :-/
> Not a problem at all. In fact, I've taken a look and it seems
> byte-optimize-form-code-walker simply ignores the value form in a
> defvar now..
> So I suspect it's commit 83983b6b7a115474572973b62eb5e42251713e63, I'm
> cc'ing Mattias, and I suspect this patch might help (but I've only
> done one quick test).
> Pip
With this patch applied to 20596a5, I still see the problems Vitalie
reported.
Joe
^ permalink raw reply [flat|nested] 13+ messages in thread
* bug#46912: 28.0.50; Byte compiled defvar with lexical binding throws (void-variable foo)
2021-03-04 23:15 ` Joseph Mingrone
@ 2021-03-05 2:52 ` Pip Cet
2021-03-05 3:22 ` Joseph Mingrone
0 siblings, 1 reply; 13+ messages in thread
From: Pip Cet @ 2021-03-05 2:52 UTC (permalink / raw)
To: Joseph Mingrone; +Cc: Mattias, Lars Ingebrigtsen, Vitalie Spinu, 46912
[-- Attachment #1: Type: text/plain, Size: 253 bytes --]
On Thu, Mar 4, 2021 at 11:15 PM Joseph Mingrone <jrm@ftfl.ca> wrote:
> With this patch applied to 20596a5, I still see the problems Vitalie
> reported.
Strange, I don't see how I managed to convince myself it works here.
Anyway, this should work.
Pip
[-- Attachment #2: 0001-Don-t-ignore-lexically-bound-variables-in-a-defvar-b.patch --]
[-- Type: text/x-patch, Size: 1286 bytes --]
From 09a195ed2ed7004c54d1701d0017a4fb10f6cbce Mon Sep 17 00:00:00 2001
From: Pip Cet <pipcet@gmail.com>
Date: Thu, 4 Mar 2021 21:06:13 +0000
Subject: [PATCH] Don't ignore lexically-bound variables in a defvar
(bug#46912)
* lisp/emacs-lisp/byte-opt.el (byte-optimize-form-code-walker): Walk
the value form of a defvar.
---
lisp/emacs-lisp/byte-opt.el | 9 ++++++---
1 file changed, 6 insertions(+), 3 deletions(-)
diff --git a/lisp/emacs-lisp/byte-opt.el b/lisp/emacs-lisp/byte-opt.el
index b51ba801552d6..b3325816c5c97 100644
--- a/lisp/emacs-lisp/byte-opt.el
+++ b/lisp/emacs-lisp/byte-opt.el
@@ -607,9 +607,12 @@ byte-optimize-form-code-walker
(setq args (cddr args)))
(cons fn (nreverse var-expr-list))))
- (`(defvar ,(and (pred symbolp) name) . ,_)
- (push name byte-optimize--dynamic-vars)
- form)
+ (`(defvar ,(and (pred symbolp) name) . ,rest)
+ (let ((optimized-rest (and rest
+ (cons (byte-optimize-form (car rest) nil)
+ (cdr rest)))))
+ (push name byte-optimize--dynamic-vars)
+ `(defvar ,name . ,optimized-rest)))
(`(,(pred byte-code-function-p) . ,exps)
(cons fn (mapcar #'byte-optimize-form exps)))
--
2.30.1
^ permalink raw reply related [flat|nested] 13+ messages in thread
* bug#46912: 28.0.50; Byte compiled defvar with lexical binding throws (void-variable foo)
2021-03-05 2:52 ` Pip Cet
@ 2021-03-05 3:22 ` Joseph Mingrone
2021-03-05 8:27 ` Vitalie Spinu
0 siblings, 1 reply; 13+ messages in thread
From: Joseph Mingrone @ 2021-03-05 3:22 UTC (permalink / raw)
To: Pip Cet; +Cc: Mattias, Lars Ingebrigtsen, Vitalie Spinu, 46912
On Fri, 2021-03-05 at 02:52, Pip Cet <pipcet@gmail.com> wrote:
> On Thu, Mar 4, 2021 at 11:15 PM Joseph Mingrone <jrm@ftfl.ca> wrote:
>> With this patch applied to 20596a5, I still see the problems Vitalie
>> reported.
> Strange, I don't see how I managed to convince myself it works here.
> Anyway, this should work.
> Pip
Hi Pip,
I can confirm that your second patch fixes the problem.
Thank you!
Joe
^ permalink raw reply [flat|nested] 13+ messages in thread
* bug#46912: 28.0.50; Byte compiled defvar with lexical binding throws (void-variable foo)
2021-03-05 3:22 ` Joseph Mingrone
@ 2021-03-05 8:27 ` Vitalie Spinu
0 siblings, 0 replies; 13+ messages in thread
From: Vitalie Spinu @ 2021-03-05 8:27 UTC (permalink / raw)
To: Joseph Mingrone; +Cc: Mattias, Lars Ingebrigtsen, Pip Cet, 46912
Hi Pip,
The second patch fixed it for me as well. Thanks!
Vitalie
^ permalink raw reply [flat|nested] 13+ messages in thread
* bug#46912: 28.0.50; Byte compiled defvar with lexical binding throws (void-variable foo)
2021-03-04 21:08 ` Pip Cet
2021-03-04 23:15 ` Joseph Mingrone
@ 2021-03-05 9:10 ` Mattias Engdegård
2021-03-05 9:56 ` Pip Cet
1 sibling, 1 reply; 13+ messages in thread
From: Mattias Engdegård @ 2021-03-05 9:10 UTC (permalink / raw)
To: Pip Cet; +Cc: Joseph Mingrone, 46912, Lars Ingebrigtsen, Vitalie Spinu
4 mars 2021 kl. 22.08 skrev Pip Cet <pipcet@gmail.com>:
> Not a problem at all. In fact, I've taken a look and it seems
> byte-optimize-form-code-walker simply ignores the value form in a
> defvar now..
Oh dear, what a silly mistake. Thank you for finding it so quickly, and that (last, corrected) patch of yours looks sound -- please push it right away.
^ permalink raw reply [flat|nested] 13+ messages in thread
* bug#46912: 28.0.50; Byte compiled defvar with lexical binding throws (void-variable foo)
2021-03-05 9:10 ` Mattias Engdegård
@ 2021-03-05 9:56 ` Pip Cet
0 siblings, 0 replies; 13+ messages in thread
From: Pip Cet @ 2021-03-05 9:56 UTC (permalink / raw)
To: Mattias Engdegård
Cc: Joseph Mingrone, Lars Ingebrigtsen, 46912-done, Vitalie Spinu
On Fri, Mar 5, 2021 at 9:11 AM Mattias Engdegård <mattiase@acm.org> wrote:
> 4 mars 2021 kl. 22.08 skrev Pip Cet <pipcet@gmail.com>:
>
> > Not a problem at all. In fact, I've taken a look and it seems
> > byte-optimize-form-code-walker simply ignores the value form in a
> > defvar now..
>
> Oh dear, what a silly mistake. Thank you for finding it so quickly, and that (last, corrected) patch of yours looks sound -- please push it right away.
Done, and closing. Sorry about the first patch, I must have "tested"
it after setting byte-optimize to nil...
Pip
^ permalink raw reply [flat|nested] 13+ messages in thread
end of thread, other threads:[~2021-03-05 9:56 UTC | newest]
Thread overview: 13+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-03-04 0:14 bug#46912: 28.0.50; Byte compiled defvar with lexical binding throws (void-variable foo) Vitalie Spinu
2021-03-04 18:03 ` Joseph Mingrone
2021-03-04 20:09 ` Lars Ingebrigtsen
2021-03-04 20:17 ` Joseph Mingrone
2021-03-04 20:48 ` Lars Ingebrigtsen
2021-03-04 21:08 ` Pip Cet
2021-03-04 23:15 ` Joseph Mingrone
2021-03-05 2:52 ` Pip Cet
2021-03-05 3:22 ` Joseph Mingrone
2021-03-05 8:27 ` Vitalie Spinu
2021-03-05 9:10 ` Mattias Engdegård
2021-03-05 9:56 ` Pip Cet
2021-03-04 20:21 ` Pip Cet
Code repositories for project(s) associated with this public inbox
https://git.savannah.gnu.org/cgit/emacs.git
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).