bug#12501: Automake generated makefiles can be less verbose when building with silent rules

[Stefano Lattarini]

tags 12501 + moreinfo
severity 12501 wishlist
thanks

On 09/24/2012 09:35 AM, Hib Eris wrote:
> Hi all,
> 
> I think Automake generated makefiles can be a little less verbose when
> building with silent rules.
> Please consider attached patch.
> 
> Thanks,
> 
> Hib Eris
>
> From b1ab343e3e5c83e3da99a34924acf9c68bd4ee72 Mon Sep 17 00:00:00 2001
> From: Hib Eris <address@hidden>
> Date: Mon, 24 Sep 2012 09:29:56 +0200
> Subject: [PATCH] Do not echo targets when building with silent rules
>
> ---
> automake.in       |    2 +-
> lib/am/subdirs.am |    2 +-
> 2 files changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/automake.in b/automake.in
> index b165855..5b383a0 100644
> --- a/automake.in
> +++ b/automake.in
> @@ -4476,7 +4476,7 @@ sub handle_all ($)
>       # things explicitly.
>       $output_all .= ("all: @local_headers"
>                       . "\n\t"
> -                     . '$(MAKE) $(AM_MAKEFLAGS) '
> +                     . '$(AM_V_at)$(MAKE) $(AM_MAKEFLAGS) '
>                       . (var ('SUBDIRS') ? 'all-recursive' : 'all-am')
>                       . "\n\n");
>         depend ('.MAKE', 'all');
> diff --git a/lib/am/subdirs.am b/lib/am/subdirs.am
> index 005483a..b430d21 100644
> --- a/lib/am/subdirs.am
> +++ b/lib/am/subdirs.am
> @@ -56,7 +56,7 @@ $(am__recursive_targets):
>         *) list='$(SUBDIRS)' ;; \
>       esac; \
>       for subdir in $$list; do \
> -       echo "Making $$target in $$subdir"; \
> +       $(AM_V_P) && echo "Making $$target in $$subdir"; \
>         if test "$$subdir" = "."; then \
>           dot_seen=yes; \
>           local_target="$$target-am"; \
> -- 1.7.5.4
>
Hmm...  I fear that this change would actually make the makefile
output *too* silent, possibly impeding debugging or problem diagnosis
in a serious way.  But I'm not sure if this is actually the case,
or if I'm just being overly wary.  Would any other automaker like
to chime in with his opinion?

Anyway, the best way to make an "informed decision" would be to try
out Automake, patched with this diff, on few real-world package (GNU
coreutils and GNU autoconf comes to mind), with GNU and non-GNU make
implementations alike, and seeing how the make output looks when
everything goes smoothly, and then how it looks when there are errors
(e.g., errors compiling a file, permission errors modifying or
creating a file, and even syntax errors in a subdir Makefile).  Would
you be willing to give this experiment a try, later posting here the
output and impressions you get?  This might be overkill for such a
minor change however ...

Waiting for feedback,
  Stefano