Make the compilation slightly less verbose?

Make the compilation slightly less verbose?

[Lars Ingebrigtsen]

At one point in the compilation process (I think it's between the first
and second pdumps?) the compilation goes very verbose all of a sudden:

make[3]: Entering directory '/home/larsi/src/emacs/trunk/lisp'
  ELC      ../lisp/dos-vars.elc
make[3]: Leaving directory '/home/larsi/src/emacs/trunk/lisp'
make[3]: Entering directory '/home/larsi/src/emacs/trunk/lisp'
  ELC      ../lisp/dos-w32.elc
make[3]: Leaving directory '/home/larsi/src/emacs/trunk/lisp'
make[3]: Entering directory '/home/larsi/src/emacs/trunk/lisp'
  ELC      ../lisp/dynamic-setting.elc
make[3]: Leaving directory '/home/larsi/src/emacs/trunk/lisp'

Would it be a good idea to get rid of that "entering"/"leaving" pair for
every single .el file that's compiled?

And if so...  how?  :-)

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: Make the compilation slightly less verbose?

[Andreas Schwab]

On Jun 14 2019, Lars Ingebrigtsen <larsi@gnus.org> wrote:

> Would it be a good idea to get rid of that "entering"/"leaving" pair for
> every single .el file that's compiled?
>
> And if so...  how?  :-)

$(MAKE) --no-print-directory ...

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510  2552 DF73 E780 A9DA AEC1
"And now for something completely different."

Re: Make the compilation slightly less verbose?

[Lars Ingebrigtsen]

Andreas Schwab <schwab@linux-m68k.org> writes:

> $(MAKE) --no-print-directory ...

Thanks!  I think I found the correct place.  With this patch:

diff --git a/src/Makefile.in b/src/Makefile.in
index be769458d3..5fc623c346 100644
--- a/src/Makefile.in
+++ b/src/Makefile.in
@@ -776,7 +776,8 @@ .PHONY:
 ## bootstrap-emacs$(EXEEXT) as an order-only prerequisite.
 
 %.elc: %.el | bootstrap-emacs$(EXEEXT) $(bootstrap_pdmp)
-	@$(MAKE) -C ../lisp EMACS="$(bootstrap_exe)" THEFILE=$< $<c
+	@$(MAKE) --no-print-directory \
+		-C ../lisp EMACS="$(bootstrap_exe)" THEFILE=$< $<c
 
 ## VCSWITNESS points to the file that holds info about the current checkout.
 ## We use it as a heuristic to decide when to rebuild loaddefs.el.

I now get

  ELC      ../lisp/dos-fns.elc
  ELC      ../lisp/dos-vars.elc
  ELC      ../lisp/dos-w32.elc
  ELC      ../lisp/dynamic-setting.elc

on a make bootstrap, and otherwise things look the same as before, I
think.

Does anybody object to this change?

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: Make the compilation slightly less verbose?

[Eli Zaretskii]

> From: Lars Ingebrigtsen <larsi@gnus.org>
> Date: Fri, 14 Jun 2019 17:27:22 +0200
> 
> At one point in the compilation process (I think it's between the first
> and second pdumps?) the compilation goes very verbose all of a sudden:

That's because we compile each of these files in a separate sub-make.

> make[3]: Entering directory '/home/larsi/src/emacs/trunk/lisp'
>   ELC      ../lisp/dos-vars.elc
> make[3]: Leaving directory '/home/larsi/src/emacs/trunk/lisp'
> make[3]: Entering directory '/home/larsi/src/emacs/trunk/lisp'
>   ELC      ../lisp/dos-w32.elc
> make[3]: Leaving directory '/home/larsi/src/emacs/trunk/lisp'
> make[3]: Entering directory '/home/larsi/src/emacs/trunk/lisp'
>   ELC      ../lisp/dynamic-setting.elc
> make[3]: Leaving directory '/home/larsi/src/emacs/trunk/lisp'
> 
> Would it be a good idea to get rid of that "entering"/"leaving" pair for
> every single .el file that's compiled?
> 
> And if so...  how?  :-)

Run "make" with the --no-print-directory switch.

Re: Make the compilation slightly less verbose?

[Andy Moreton]

On Fri 14 Jun 2019, Lars Ingebrigtsen wrote:

> At one point in the compilation process (I think it's between the first
> and second pdumps?) the compilation goes very verbose all of a sudden:
>
> make[3]: Entering directory '/home/larsi/src/emacs/trunk/lisp'
>   ELC      ../lisp/dos-vars.elc
> make[3]: Leaving directory '/home/larsi/src/emacs/trunk/lisp'
> make[3]: Entering directory '/home/larsi/src/emacs/trunk/lisp'
>   ELC      ../lisp/dos-w32.elc
> make[3]: Leaving directory '/home/larsi/src/emacs/trunk/lisp'
> make[3]: Entering directory '/home/larsi/src/emacs/trunk/lisp'
>   ELC      ../lisp/dynamic-setting.elc
> make[3]: Leaving directory '/home/larsi/src/emacs/trunk/lisp'
>
> Would it be a good idea to get rid of that "entering"/"leaving" pair for
> every single .el file that's compiled?
>
> And if so...  how?  :-)

Unrelated, to the above, but you recently added this:
"master 1d2e4d2: Silence okuri-nasi count-up messaging during
compilation"

This is useful when bootstrapping, as that stage takes ages, so it is
helpful to indicate that it has not crashed and is making forward
progress.

Please consider reverting that change.

    AndyM

Re: Make the compilation slightly less verbose?

[Eli Zaretskii]

> From: Lars Ingebrigtsen <larsi@gnus.org>
> Date: Fri, 14 Jun 2019 18:10:22 +0200
> Cc: emacs-devel@gnu.org
> 
> diff --git a/src/Makefile.in b/src/Makefile.in
> index be769458d3..5fc623c346 100644
> --- a/src/Makefile.in
> +++ b/src/Makefile.in
> @@ -776,7 +776,8 @@ .PHONY:
>  ## bootstrap-emacs$(EXEEXT) as an order-only prerequisite.
>  
>  %.elc: %.el | bootstrap-emacs$(EXEEXT) $(bootstrap_pdmp)
> -	@$(MAKE) -C ../lisp EMACS="$(bootstrap_exe)" THEFILE=$< $<c
> +	@$(MAKE) --no-print-directory \
> +		-C ../lisp EMACS="$(bootstrap_exe)" THEFILE=$< $<c
>  
>  ## VCSWITNESS points to the file that holds info about the current checkout.
>  ## We use it as a heuristic to decide when to rebuild loaddefs.el.
> 
> I now get
> 
>   ELC      ../lisp/dos-fns.elc
>   ELC      ../lisp/dos-vars.elc
>   ELC      ../lisp/dos-w32.elc
>   ELC      ../lisp/dynamic-setting.elc
> 
> on a make bootstrap, and otherwise things look the same as before, I
> think.
> 
> Does anybody object to this change?

I don't really see why your personal preferences should be forced on
all of us.

Re: Make the compilation slightly less verbose?

[Alan Mackenzie]

Hello, Eli.

On Fri, Jun 14, 2019 at 21:58:04 +0300, Eli Zaretskii wrote:
> > From: Lars Ingebrigtsen <larsi@gnus.org>
> > Date: Fri, 14 Jun 2019 18:10:22 +0200
> > Cc: emacs-devel@gnu.org

> > diff --git a/src/Makefile.in b/src/Makefile.in
> > index be769458d3..5fc623c346 100644
> > --- a/src/Makefile.in
> > +++ b/src/Makefile.in
> > @@ -776,7 +776,8 @@ .PHONY:
> >  ## bootstrap-emacs$(EXEEXT) as an order-only prerequisite.

> >  %.elc: %.el | bootstrap-emacs$(EXEEXT) $(bootstrap_pdmp)
> > -	@$(MAKE) -C ../lisp EMACS="$(bootstrap_exe)" THEFILE=$< $<c
> > +	@$(MAKE) --no-print-directory \
> > +		-C ../lisp EMACS="$(bootstrap_exe)" THEFILE=$< $<c

> >  ## VCSWITNESS points to the file that holds info about the current checkout.
> >  ## We use it as a heuristic to decide when to rebuild loaddefs.el.

> > I now get

> >   ELC      ../lisp/dos-fns.elc
> >   ELC      ../lisp/dos-vars.elc
> >   ELC      ../lisp/dos-w32.elc
> >   ELC      ../lisp/dynamic-setting.elc

> > on a make bootstrap, and otherwise things look the same as before, I
> > think.

> > Does anybody object to this change?

> I don't really see why your personal preferences should be forced on
> all of us.

Well, somebody's personal preference must prevail, unless we introduce a
new configuration option (PLEASE NOT!).

If I understand this correctly, three lines of output were being output
for each file, and the information content of two of these was close to,
if not actually, zero.

So I see this as an objectively desirable change.

-- 
Alan Mackenzie (Nuremberg, Germany).

Re: Make the compilation slightly less verbose?

[Óscar Fuentes]

Eli Zaretskii <eliz@gnu.org> writes:

>> Does anybody object to this change?
>
> I don't really see why your personal preferences should be forced on
> all of us.

He is asking.

Re: Make the compilation slightly less verbose?

[Werner LEMBERG]

>> >   ELC      ../lisp/dos-fns.elc
>> >   ELC      ../lisp/dos-vars.elc
>> >   ELC      ../lisp/dos-w32.elc
>> >   ELC      ../lisp/dynamic-setting.elc
> 
>> I don't really see why your personal preferences should be forced on
>> all of us.
> 
> If I understand this correctly, three lines of output were being
> output for each file, and the information content of two of these
> was close to, if not actually, zero.
> 
> So I see this as an objectively desirable change.

+1


    Werner

Re: Make the compilation slightly less verbose?

[David Ringo]

On Fri, Jun 14, 2019 at 10:56 AM Lars Ingebrigtsen <larsi@gnus.org> wrote:
>
> Andreas Schwab <schwab@linux-m68k.org> writes:
>
> > $(MAKE) --no-print-directory ...
>
> Thanks!  I think I found the correct place.  With this patch:
>
> diff --git a/src/Makefile.in b/src/Makefile.in
> index be769458d3..5fc623c346 100644
> --- a/src/Makefile.in
> +++ b/src/Makefile.in
> @@ -776,7 +776,8 @@ .PHONY:
>  ## bootstrap-emacs$(EXEEXT) as an order-only prerequisite.
>
>  %.elc: %.el | bootstrap-emacs$(EXEEXT) $(bootstrap_pdmp)
> -       @$(MAKE) -C ../lisp EMACS="$(bootstrap_exe)" THEFILE=$< $<c
> +       @$(MAKE) --no-print-directory \
> +               -C ../lisp EMACS="$(bootstrap_exe)" THEFILE=$< $<c
>
>  ## VCSWITNESS points to the file that holds info about the current checkout.
>  ## We use it as a heuristic to decide when to rebuild loaddefs.el.
>
> I now get
>
>   ELC      ../lisp/dos-fns.elc
>   ELC      ../lisp/dos-vars.elc
>   ELC      ../lisp/dos-w32.elc
>   ELC      ../lisp/dynamic-setting.elc
>
> on a make bootstrap, and otherwise things look the same as before, I
> think.
>
> Does anybody object to this change?
>
> --
> (domestic pets only, the antidote for overdose, milk.)
>    bloggy blog: http://lars.ingebrigtsen.no
>

It's worth noting that you can add --no-print-directory to
GNUMAKEFLAGS in the environment to affect *all* recursive invocations.
I don't know if that's generally desirable for you though.

Re: Make the compilation slightly less verbose?

[Stefan Monnier]

>>  %.elc: %.el | bootstrap-emacs$(EXEEXT) $(bootstrap_pdmp)
>> -	@$(MAKE) -C ../lisp EMACS="$(bootstrap_exe)" THEFILE=$< $<c
>> +	@$(MAKE) --no-print-directory \
>> +		-C ../lisp EMACS="$(bootstrap_exe)" THEFILE=$< $<c
[...]
>> Does anybody object to this change?
> I don't really see why your personal preferences should be forced on
> all of us.

FWIW, I think it's an improvement.  But I'll mention that I'd prefer if
we got this kind of output by eliminating the intermediate `make` ;-)


        Stefan

Re: Make the compilation slightly less verbose?

[Paul Eggert]

On 6/14/19 1:14 PM, Stefan Monnier wrote:
> FWIW, I think it's an improvement.  But I'll mention that I'd prefer if
> we got this kind of output by eliminating the intermediate `make` ;-)

Likewise, on both counts.

Re: Make the compilation slightly less verbose?

[Eli Zaretskii]

> Date: Fri, 14 Jun 2019 19:28:49 +0000
> Cc: Lars Ingebrigtsen <larsi@gnus.org>, schwab@linux-m68k.org,
>   emacs-devel@gnu.org
> From: Alan Mackenzie <acm@muc.de>
> 
> If I understand this correctly, three lines of output were being output
> for each file, and the information content of two of these was close to,
> if not actually, zero.
> 
> So I see this as an objectively desirable change.

I'm saying that it's very easy to invoke Make with the relevant
option, if you don't like this output.  I definitely remember a few
situations where this very output helped me debug some user's problems
with the build, so making this the default will make such debugging a
tad more difficult.

That said, I'm not going to fight this change.

Re: Make the compilation slightly less verbose?

[Eli Zaretskii]

> Date: Sat, 15 Jun 2019 09:00:14 +0300
> From: Eli Zaretskii <eliz@gnu.org>
> Cc: larsi@gnus.org, schwab@linux-m68k.org, emacs-devel@gnu.org
> 
> I'm saying that it's very easy to invoke Make with the relevant
> option, if you don't like this output.  I definitely remember a few
> situations where this very output helped me debug some user's problems
> with the build, so making this the default will make such debugging a
> tad more difficult.
> 
> That said, I'm not going to fight this change.

Please make the change be conditioned on the verbosity, i.e. I'd very
much like it if "make V=1" would still emit these messages.

Thanks.

Re: Make the compilation slightly less verbose?

[Stefan Monnier]

> Please make the change be conditioned on the verbosity, i.e. I'd very
> much like it if "make V=1" would still emit these messages.

+1


        Stefan

Re: Make the compilation slightly less verbose?

[Lars Ingebrigtsen]

Eli Zaretskii <eliz@gnu.org> writes:

> Please make the change be conditioned on the verbosity, i.e. I'd very
> much like it if "make V=1" would still emit these messages.

I seem to have almost completely successfully managed to suppress all my
knowledge I may once have had of how autoconf works, but...  Does this
look right?

diff --git a/src/Makefile.in b/src/Makefile.in
index be769458d3..b89a2f890e 100644
--- a/src/Makefile.in
+++ b/src/Makefile.in
@@ -776,7 +776,12 @@ .PHONY:
 ## bootstrap-emacs$(EXEEXT) as an order-only prerequisite.
 
 %.elc: %.el | bootstrap-emacs$(EXEEXT) $(bootstrap_pdmp)
+ifeq (@AM_V@,)
+	@$(MAKE) --no-print-directory\
+		-C ../lisp EMACS="$(bootstrap_exe)" THEFILE=$< $<c
+else
 	@$(MAKE) -C ../lisp EMACS="$(bootstrap_exe)" THEFILE=$< $<c
+endif
 
 ## VCSWITNESS points to the file that holds info about the current checkout.
 ## We use it as a heuristic to decide when to rebuild loaddefs.el.


$ V=1 make bootstrap

now gives me:

make[3]: Entering directory '/home/larsi/src/emacs/trunk/lisp'
EMACSLOADPATH= '../src/bootstrap-emacs' -batch --no-site-file --no-site-lisp --eval '(setq load-prefer-newer t)'  \
	-l bytecomp -f byte-compile-refresh-preloaded \
	-f batch-byte-compile ../lisp/dos-w32.el
make[3]: Leaving directory '/home/larsi/src/emacs/trunk/lisp'
make[3]: Entering directory '/home/larsi/src/emacs/trunk/lisp'
EMACSLOADPATH= '../src/bootstrap-emacs' -batch --no-site-file --no-site-lisp --eval '(setq load-prefer-newer t)'  \
	-l bytecomp -f byte-compile-refresh-preloaded \
	-f batch-byte-compile ../lisp/dynamic-setting.el
make[3]: Leaving directory '/home/larsi/src/emacs/trunk/lisp'

which is what it's supposed to be, I think?  Just "make bootstrap" gives
me:

  ELC      ../lisp/dos-w32.elc
  ELC      ../lisp/dynamic-setting.elc

so I think it works, but whether that's idiomatic use of the autoconf
machinery, I'm not sure...

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: Make the compilation slightly less verbose?

[Lars Ingebrigtsen]

David Ringo <davidmringo@gmail.com> writes:

> It's worth noting that you can add --no-print-directory to
> GNUMAKEFLAGS in the environment to affect *all* recursive invocations.
> I don't know if that's generally desirable for you though.

No, the rest of the directory printing seems useful and informative to
me.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: Make the compilation slightly less verbose?

[Lars Ingebrigtsen]

Andy Moreton <andrewjmoreton@gmail.com> writes:

> Unrelated, to the above, but you recently added this:
> "master 1d2e4d2: Silence okuri-nasi count-up messaging during
> compilation"
>
> This is useful when bootstrapping, as that stage takes ages, so it is
> helpful to indicate that it has not crashed and is making forward
> progress.
>
> Please consider reverting that change.

Compiling that file didn't seem to take particularly long for me, so the
messaging seemed odd to me, but as it's apparently an issue for others,
I've reverted the change.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: Make the compilation slightly less verbose?

[Eli Zaretskii]

> From: Lars Ingebrigtsen <larsi@gnus.org>
> Cc: acm@muc.de,  schwab@linux-m68k.org,  emacs-devel@gnu.org
> Date: Sat, 15 Jun 2019 15:29:39 +0200
> 
> diff --git a/src/Makefile.in b/src/Makefile.in
> index be769458d3..b89a2f890e 100644
> --- a/src/Makefile.in
> +++ b/src/Makefile.in
> @@ -776,7 +776,12 @@ .PHONY:
>  ## bootstrap-emacs$(EXEEXT) as an order-only prerequisite.
>  
>  %.elc: %.el | bootstrap-emacs$(EXEEXT) $(bootstrap_pdmp)
> +ifeq (@AM_V@,)
> +	@$(MAKE) --no-print-directory\
> +		-C ../lisp EMACS="$(bootstrap_exe)" THEFILE=$< $<c
> +else
>  	@$(MAKE) -C ../lisp EMACS="$(bootstrap_exe)" THEFILE=$< $<c
> +endif

I think this should depend on V being equal or unequal to 0.  Because
one should be able to say "make V=0" and still have these messages
suppressed.

> $ V=1 make bootstrap
> 
> now gives me:
> 
> make[3]: Entering directory '/home/larsi/src/emacs/trunk/lisp'
> EMACSLOADPATH= '../src/bootstrap-emacs' -batch --no-site-file --no-site-lisp --eval '(setq load-prefer-newer t)'  \
> 	-l bytecomp -f byte-compile-refresh-preloaded \
> 	-f batch-byte-compile ../lisp/dos-w32.el
> make[3]: Leaving directory '/home/larsi/src/emacs/trunk/lisp'
> make[3]: Entering directory '/home/larsi/src/emacs/trunk/lisp'
> EMACSLOADPATH= '../src/bootstrap-emacs' -batch --no-site-file --no-site-lisp --eval '(setq load-prefer-newer t)'  \
> 	-l bytecomp -f byte-compile-refresh-preloaded \
> 	-f batch-byte-compile ../lisp/dynamic-setting.el
> make[3]: Leaving directory '/home/larsi/src/emacs/trunk/lisp'
> 
> which is what it's supposed to be, I think?

Yes, but what happens if you say "V=0 make bootstrap"?

Re: Make the compilation slightly less verbose?

[Lars Ingebrigtsen]

Eli Zaretskii <eliz@gnu.org> writes:

> I think this should depend on V being equal or unequal to 0.  Because
> one should be able to say "make V=0" and still have these messages
> suppressed.

Ah, is that how the V variable is supposed to work?

That is...  what values of V should make the process verbose?
Everything but "" and 0, or is it just 1 that should make it verbose?

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: Make the compilation slightly less verbose?

[Eli Zaretskii]

> From: Lars Ingebrigtsen <larsi@gnus.org>
> Cc: acm@muc.de,  schwab@linux-m68k.org,  emacs-devel@gnu.org
> Date: Sat, 15 Jun 2019 15:55:29 +0200
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > I think this should depend on V being equal or unequal to 0.  Because
> > one should be able to say "make V=0" and still have these messages
> > suppressed.
> 
> Ah, is that how the V variable is supposed to work?
> 
> That is...  what values of V should make the process verbose?
> Everything but "" and 0, or is it just 1 that should make it verbose?

I think only 0 and 1 have defined behavior.

Re: Make the compilation slightly less verbose?

[Lars Ingebrigtsen]

Eli Zaretskii <eliz@gnu.org> writes:

> I think only 0 and 1 have defined behavior.

OK, I'll change the test and check it in, and then perhaps some autoconf
expert can tweak it if there's a more idiomatic way to do the test (by
defining another macro or something)...

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: Make the compilation slightly less verbose?

[Andreas Schwab]

On Jun 15 2019, Lars Ingebrigtsen <larsi@gnus.org> wrote:

> Eli Zaretskii <eliz@gnu.org> writes:
>
>> Please make the change be conditioned on the verbosity, i.e. I'd very
>> much like it if "make V=1" would still emit these messages.
>
> I seem to have almost completely successfully managed to suppress all my
> knowledge I may once have had of how autoconf works, but...  Does this
> look right?
>
> diff --git a/src/Makefile.in b/src/Makefile.in
> index be769458d3..b89a2f890e 100644
> --- a/src/Makefile.in
> +++ b/src/Makefile.in
> @@ -776,7 +776,12 @@ .PHONY:
>  ## bootstrap-emacs$(EXEEXT) as an order-only prerequisite.
>  
>  %.elc: %.el | bootstrap-emacs$(EXEEXT) $(bootstrap_pdmp)
> +ifeq (@AM_V@,)
> +	@$(MAKE) --no-print-directory\
> +		-C ../lisp EMACS="$(bootstrap_exe)" THEFILE=$< $<c
> +else
>  	@$(MAKE) -C ../lisp EMACS="$(bootstrap_exe)" THEFILE=$< $<c
> +endif

AM_V_NO_PD = $(am__v_NO_PD_@AM_V@)
am__v_NO_PD_ = $(am__v_NO_PD_@AM_DEFAULT_V@)
am__v_NO_PD_0 = --no-print-directory
am__v_NO_PD_1 =

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510  2552 DF73 E780 A9DA AEC1
"And now for something completely different."

Re: Make the compilation slightly less verbose?

[Lars Ingebrigtsen]

Andreas Schwab <schwab@linux-m68k.org> writes:

> AM_V_NO_PD = $(am__v_NO_PD_@AM_V@)
> am__v_NO_PD_ = $(am__v_NO_PD_@AM_DEFAULT_V@)
> am__v_NO_PD_0 = --no-print-directory
> am__v_NO_PD_1 =

Thanks; that looks much more idiomatic.  I'll push shortly.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no