* Define progn_ignore to run Fprogn and ignore output
@ 2016-12-27 2:14 Chris Gregory
2016-12-27 6:26 ` Eli Zaretskii
2016-12-27 15:49 ` Alan Mackenzie
0 siblings, 2 replies; 8+ messages in thread
From: Chris Gregory @ 2016-12-27 2:14 UTC (permalink / raw)
To: emacs-devel
This patch defines progn_ignore to do Fprogn but not return a result.
This will minorly speed up all operations using `unwind_body'. It also
simplifies the code for `Fprog1' without sacrificing performance.
diff --git a/src/eval.c b/src/eval.c
index ddcccc2..7e02838 100644
--- a/src/eval.c
+++ b/src/eval.c
@@ -456,10 +456,17 @@ usage: (progn BODY...) */)
/* Evaluate BODY sequentially, discarding its value. Suitable for
record_unwind_protect. */
+static void
+progn_ignore (Lisp_Object body)
+{
+ while (CONSP (body))
+ eval_sub (XCAR (body));
+}
+
void
unwind_body (Lisp_Object body)
{
- Fprogn (body);
+ progn_ignore (body);
}
DEFUN ("prog1", Fprog1, Sprog1, 1, UNEVALLED, 0,
@@ -470,15 +477,8 @@ usage: (prog1 FIRST BODY...) */)
(Lisp_Object args)
{
Lisp_Object val;
- Lisp_Object args_left;
-
- args_left = args;
- val = args;
-
- val = eval_sub (XCAR (args_left));
- while (CONSP (args_left = XCDR (args_left)))
- eval_sub (XCAR (args_left));
-
+ val = eval_sub (XCAR (args));
+ progn_ignore (XCDR (args));
return val;
}
^ permalink raw reply related [flat|nested] 8+ messages in thread
* Re: Define progn_ignore to run Fprogn and ignore output
2016-12-27 2:14 Chris Gregory
@ 2016-12-27 6:26 ` Eli Zaretskii
2016-12-27 16:28 ` Chris Gregory
2016-12-27 15:49 ` Alan Mackenzie
1 sibling, 1 reply; 8+ messages in thread
From: Eli Zaretskii @ 2016-12-27 6:26 UTC (permalink / raw)
To: Chris Gregory; +Cc: emacs-devel
> From: Chris Gregory <czipperz@gmail.com>
> Date: Mon, 26 Dec 2016 20:14:47 -0600
>
> This patch defines progn_ignore to do Fprogn but not return a result.
> This will minorly speed up all operations using `unwind_body'.
Thanks. Can you give some numbers that show what kind of speed-up
this will yield?
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: Define progn_ignore to run Fprogn and ignore output
2016-12-27 2:14 Chris Gregory
2016-12-27 6:26 ` Eli Zaretskii
@ 2016-12-27 15:49 ` Alan Mackenzie
2016-12-27 16:28 ` Chris Gregory
1 sibling, 1 reply; 8+ messages in thread
From: Alan Mackenzie @ 2016-12-27 15:49 UTC (permalink / raw)
To: Chris Gregory; +Cc: emacs-devel
Hello, Chris.
Just a small point about your proposed patch.
On Mon, Dec 26, 2016 at 08:14:47PM -0600, Chris Gregory wrote:
> This patch defines progn_ignore to do Fprogn but not return a result.
> This will minorly speed up all operations using `unwind_body'. It also
> simplifies the code for `Fprog1' without sacrificing performance.
> diff --git a/src/eval.c b/src/eval.c
> index ddcccc2..7e02838 100644
> --- a/src/eval.c
> +++ b/src/eval.c
> @@ -456,10 +456,17 @@ usage: (progn BODY...) */)
> /* Evaluate BODY sequentially, discarding its value. Suitable for
> record_unwind_protect. */
> +static void
> +progn_ignore (Lisp_Object body)
> +{
> + while (CONSP (body))
> + eval_sub (XCAR (body));
> +}
> +
Doesn't the above code up an infinite loop? `body' appears not to be
changed between iterations of the while.
> void
> unwind_body (Lisp_Object body)
> {
> - Fprogn (body);
> + progn_ignore (body);
> }
> DEFUN ("prog1", Fprog1, Sprog1, 1, UNEVALLED, 0,
> @@ -470,15 +477,8 @@ usage: (prog1 FIRST BODY...) */)
> (Lisp_Object args)
> {
> Lisp_Object val;
> - Lisp_Object args_left;
> -
> - args_left = args;
> - val = args;
> -
> - val = eval_sub (XCAR (args_left));
> - while (CONSP (args_left = XCDR (args_left)))
> - eval_sub (XCAR (args_left));
> -
> + val = eval_sub (XCAR (args));
> + progn_ignore (XCDR (args));
> return val;
> }
--
Alan Mackenzie (Nuremberg, Germany).
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: Define progn_ignore to run Fprogn and ignore output
2016-12-27 6:26 ` Eli Zaretskii
@ 2016-12-27 16:28 ` Chris Gregory
2016-12-27 18:45 ` Paul Eggert
0 siblings, 1 reply; 8+ messages in thread
From: Chris Gregory @ 2016-12-27 16:28 UTC (permalink / raw)
To: Eli Zaretskii; +Cc: Emacs Development Mailing List
[-- Attachment #1: Type: text/plain, Size: 721 bytes --]
It was just a guess to be perfectly honest. I'll try to benchmark it
tomorrow. If it inlines the function to Fprogn then there is no benefit to
performance (assuming it elides the Lisp_Object completely). The code,
however, is cleaner now. I'll look through the rest of the code for Fprogn
usages.
Sincerely,
Chris Gregory
On Tue, Dec 27, 2016 at 12:26 AM, Eli Zaretskii <eliz@gnu.org> wrote:
> > From: Chris Gregory <czipperz@gmail.com>
> > Date: Mon, 26 Dec 2016 20:14:47 -0600
> >
> > This patch defines progn_ignore to do Fprogn but not return a result.
> > This will minorly speed up all operations using `unwind_body'.
>
> Thanks. Can you give some numbers that show what kind of speed-up
> this will yield?
>
[-- Attachment #2: Type: text/html, Size: 1415 bytes --]
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: Define progn_ignore to run Fprogn and ignore output
2016-12-27 15:49 ` Alan Mackenzie
@ 2016-12-27 16:28 ` Chris Gregory
0 siblings, 0 replies; 8+ messages in thread
From: Chris Gregory @ 2016-12-27 16:28 UTC (permalink / raw)
To: Alan Mackenzie; +Cc: Emacs Development Mailing List
[-- Attachment #1: Type: text/plain, Size: 1677 bytes --]
Oh yes you are correct. That's my bad.
Sincerely,
Chris Gregory
On Tue, Dec 27, 2016 at 9:49 AM, Alan Mackenzie <acm@muc.de> wrote:
> Hello, Chris.
>
> Just a small point about your proposed patch.
>
> On Mon, Dec 26, 2016 at 08:14:47PM -0600, Chris Gregory wrote:
> > This patch defines progn_ignore to do Fprogn but not return a result.
> > This will minorly speed up all operations using `unwind_body'. It also
> > simplifies the code for `Fprog1' without sacrificing performance.
>
> > diff --git a/src/eval.c b/src/eval.c
> > index ddcccc2..7e02838 100644
> > --- a/src/eval.c
> > +++ b/src/eval.c
> > @@ -456,10 +456,17 @@ usage: (progn BODY...) */)
> > /* Evaluate BODY sequentially, discarding its value. Suitable for
> > record_unwind_protect. */
>
> > +static void
> > +progn_ignore (Lisp_Object body)
> > +{
> > + while (CONSP (body))
> > + eval_sub (XCAR (body));
> > +}
> > +
>
> Doesn't the above code up an infinite loop? `body' appears not to be
> changed between iterations of the while.
>
> > void
> > unwind_body (Lisp_Object body)
> > {
> > - Fprogn (body);
> > + progn_ignore (body);
> > }
>
> > DEFUN ("prog1", Fprog1, Sprog1, 1, UNEVALLED, 0,
> > @@ -470,15 +477,8 @@ usage: (prog1 FIRST BODY...) */)
> > (Lisp_Object args)
> > {
> > Lisp_Object val;
> > - Lisp_Object args_left;
> > -
> > - args_left = args;
> > - val = args;
> > -
> > - val = eval_sub (XCAR (args_left));
> > - while (CONSP (args_left = XCDR (args_left)))
> > - eval_sub (XCAR (args_left));
> > -
> > + val = eval_sub (XCAR (args));
> > + progn_ignore (XCDR (args));
> > return val;
> > }
>
> --
> Alan Mackenzie (Nuremberg, Germany).
>
[-- Attachment #2: Type: text/html, Size: 2803 bytes --]
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: Define progn_ignore to run Fprogn and ignore output
2016-12-27 16:28 ` Chris Gregory
@ 2016-12-27 18:45 ` Paul Eggert
0 siblings, 0 replies; 8+ messages in thread
From: Paul Eggert @ 2016-12-27 18:45 UTC (permalink / raw)
To: Chris Gregory; +Cc: emacs-devel
[-- Attachment #1: Type: text/plain, Size: 514 bytes --]
Chris Gregory wrote:
> I'll try to benchmark it
> tomorrow. If it inlines the function to Fprogn then there is no benefit to
> performance (assuming it elides the Lisp_Object completely). The code,
> however, is cleaner now.
Thanks, I looked into this and there's no need to benchmark, as the change (when
fixed up) does not alter the generated machine code, at least on my platform
(Fedora 24 x86-64, GCC 6.3.1). The basic idea does simplify and clarify the code
a bit, so I installed the attached.
[-- Attachment #2: 0001-Simplify-prog1-implementation.txt --]
[-- Type: text/plain, Size: 3243 bytes --]
From a02ca7a231c3856efd57a502c6a73e6c251091e8 Mon Sep 17 00:00:00 2001
From: Paul Eggert <eggert@cs.ucla.edu>
Date: Tue, 27 Dec 2016 10:32:44 -0800
Subject: [PATCH] Simplify prog1 implementation
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Inspired by a suggestion from Chris Gregory in:
http://lists.gnu.org/archive/html/emacs-devel/2016-12/msg00965.html
On my platform, this generates exactly the same machine insns.
* src/eval.c (prog_ignore): Rename from unwind_body, since
it’s more general than that. All callers changed.
(Fprog1): Simplify by using prog_ignore.
(Fwhile): Clarify by using prog_ignore.
---
src/bytecode.c | 2 +-
src/eval.c | 21 ++++++---------------
src/lisp.h | 2 +-
3 files changed, 8 insertions(+), 17 deletions(-)
diff --git a/src/bytecode.c b/src/bytecode.c
index d484dbb..3bb96c2 100644
--- a/src/bytecode.c
+++ b/src/bytecode.c
@@ -809,7 +809,7 @@ exec_byte_code (Lisp_Object bytestr, Lisp_Object vector, Lisp_Object maxdepth,
{
Lisp_Object handler = POP;
/* Support for a function here is new in 24.4. */
- record_unwind_protect (FUNCTIONP (handler) ? bcall0 : unwind_body,
+ record_unwind_protect (FUNCTIONP (handler) ? bcall0 : prog_ignore,
handler);
NEXT;
}
diff --git a/src/eval.c b/src/eval.c
index ddcccc2..e50e26a 100644
--- a/src/eval.c
+++ b/src/eval.c
@@ -453,11 +453,10 @@ usage: (progn BODY...) */)
return val;
}
-/* Evaluate BODY sequentially, discarding its value. Suitable for
- record_unwind_protect. */
+/* Evaluate BODY sequentially, discarding its value. */
void
-unwind_body (Lisp_Object body)
+prog_ignore (Lisp_Object body)
{
Fprogn (body);
}
@@ -469,16 +468,8 @@ whose values are discarded.
usage: (prog1 FIRST BODY...) */)
(Lisp_Object args)
{
- Lisp_Object val;
- Lisp_Object args_left;
-
- args_left = args;
- val = args;
-
- val = eval_sub (XCAR (args_left));
- while (CONSP (args_left = XCDR (args_left)))
- eval_sub (XCAR (args_left));
-
+ Lisp_Object val = eval_sub (XCAR (args));
+ prog_ignore (XCDR (args));
return val;
}
@@ -988,7 +979,7 @@ usage: (while TEST BODY...) */)
while (!NILP (eval_sub (test)))
{
QUIT;
- Fprogn (body);
+ prog_ignore (body);
}
return Qnil;
@@ -1191,7 +1182,7 @@ usage: (unwind-protect BODYFORM UNWINDFORMS...) */)
Lisp_Object val;
ptrdiff_t count = SPECPDL_INDEX ();
- record_unwind_protect (unwind_body, XCDR (args));
+ record_unwind_protect (prog_ignore, XCDR (args));
val = eval_sub (XCAR (args));
return unbind_to (count, val);
}
diff --git a/src/lisp.h b/src/lisp.h
index dc2c7a6..1a586ca 100644
--- a/src/lisp.h
+++ b/src/lisp.h
@@ -3867,7 +3867,7 @@ extern Lisp_Object safe_call1 (Lisp_Object, Lisp_Object);
extern Lisp_Object safe_call2 (Lisp_Object, Lisp_Object, Lisp_Object);
extern void init_eval (void);
extern void syms_of_eval (void);
-extern void unwind_body (Lisp_Object);
+extern void prog_ignore (Lisp_Object);
extern ptrdiff_t record_in_backtrace (Lisp_Object, Lisp_Object *, ptrdiff_t);
extern void mark_specpdl (union specbinding *first, union specbinding *ptr);
extern void get_backtrace (Lisp_Object array);
--
2.7.4
^ permalink raw reply related [flat|nested] 8+ messages in thread
* Re: Define progn_ignore to run Fprogn and ignore output
@ 2016-12-28 1:30 Chris Gregory
2016-12-28 6:26 ` Paul Eggert
0 siblings, 1 reply; 8+ messages in thread
From: Chris Gregory @ 2016-12-28 1:30 UTC (permalink / raw)
Your implementation is not what I intended. Please add the following
diff. I also fixed an additional place that still was declaring
unwind_body.
diff --git a/src/eval.c b/src/eval.c
index e50e26a..abd4028 100644
--- a/src/eval.c
+++ b/src/eval.c
@@ -458,7 +458,11 @@ usage: (progn BODY...) */)
void
prog_ignore (Lisp_Object body)
{
- Fprogn (body);
+ while (CONSP (body))
+ {
+ eval_sub (XCAR (body));
+ body = XCDR (body);
+ }
}
DEFUN ("prog1", Fprog1, Sprog1, 1, UNEVALLED, 0,
diff --git a/test/manual/etags/c-src/emacs/src/lisp.h b/test/manual/etags/c-src/emacs/src/lisp.h
index db87229..1b9d82b 100644
--- a/test/manual/etags/c-src/emacs/src/lisp.h
+++ b/test/manual/etags/c-src/emacs/src/lisp.h
@@ -4042,7 +4042,7 @@ extern Lisp_Object safe_call1 (Lisp_Object, Lisp_Object);
extern Lisp_Object safe_call2 (Lisp_Object, Lisp_Object, Lisp_Object);
extern void init_eval (void);
extern void syms_of_eval (void);
-extern void unwind_body (Lisp_Object);
+extern void prog_ignore (Lisp_Object);
extern ptrdiff_t record_in_backtrace (Lisp_Object, Lisp_Object *, ptrdiff_t);
extern void mark_specpdl (void);
extern void get_backtrace (Lisp_Object array);
^ permalink raw reply related [flat|nested] 8+ messages in thread
* Re: Define progn_ignore to run Fprogn and ignore output
2016-12-28 1:30 Define progn_ignore to run Fprogn and ignore output Chris Gregory
@ 2016-12-28 6:26 ` Paul Eggert
0 siblings, 0 replies; 8+ messages in thread
From: Paul Eggert @ 2016-12-28 6:26 UTC (permalink / raw)
To: Chris Gregory, emacs-devel
Chris Gregory wrote:
> Your implementation is not what I intended. Please add the following
> diff.
That diff makes the source code harder to read and understand, and the resulting
machine code is no more efficient. So I don't see why we should install the diff.
> I also fixed an additional place
The master branch already had that fix.
^ permalink raw reply [flat|nested] 8+ messages in thread
end of thread, other threads:[~2016-12-28 6:26 UTC | newest]
Thread overview: 8+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2016-12-28 1:30 Define progn_ignore to run Fprogn and ignore output Chris Gregory
2016-12-28 6:26 ` Paul Eggert
-- strict thread matches above, loose matches on Subject: below --
2016-12-27 2:14 Chris Gregory
2016-12-27 6:26 ` Eli Zaretskii
2016-12-27 16:28 ` Chris Gregory
2016-12-27 18:45 ` Paul Eggert
2016-12-27 15:49 ` Alan Mackenzie
2016-12-27 16:28 ` Chris Gregory
Code repositories for project(s) associated with this external index
https://git.savannah.gnu.org/cgit/emacs.git
https://git.savannah.gnu.org/cgit/emacs/org-mode.git
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.