GNU bug report logs - #79049
ChangeLog is added to distribution even when inside a false conditional

Previous Next

To reply to this bug, email your comments to 79049 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: Tomas Volf <~@wolfsden.cz>
To: bug-automake <at> gnu.org
Subject: ChangeLog is added to distribution even when inside a false
 conditional
Date: Sat, 19 Jul 2025 15:59:20 +0200

Hello,

when you have a rule to generate ChangeLog inside a conditional, the
file is always added to am__DIST_COMMON, even when the conditional is
false.  This leads to a failure to build the tarball, since the
ChangeLog file nor the rule to build it exist.  I suspect it might be a
bug.

Following two files are enough to reproduce it:

configure.ac:
--8<---------------cut here---------------start------------->8---
AC_PREREQ([2.71])
AC_INIT([foo], [1])
AM_INIT_AUTOMAKE([foreign])

AM_CONDITIONAL([DISTRIBUTE_CHANGELOG], [false])

AC_CONFIG_FILES([Makefile])
AC_OUTPUT
--8<---------------cut here---------------end--------------->8---

Makefile.am:
--8<---------------cut here---------------start------------->8---
if DISTRIBUTE_CHANGELOG
ChangeLog:
	echo nop
endif
--8<---------------cut here---------------end--------------->8---

When I try to build a distribution tarball out of this, I am told:

--8<---------------cut here---------------start------------->8---
$ autoreconf -vfi . && ./configure && make dist
autoreconf: export WARNINGS=
autoreconf: Entering directory '.'
autoreconf: configure.ac: not using Gettext
autoreconf: running: aclocal --force 
autoreconf: configure.ac: tracing
autoreconf: configure.ac: not using Libtool
autoreconf: configure.ac: not using Intltool
autoreconf: configure.ac: not using Gtkdoc
autoreconf: running: /gnu/store/9gzl76xq8pz6hdrgai6ydnvmi13226xh-autoconf-2.71/bin/autoconf --force
autoreconf: configure.ac: not using Autoheader
autoreconf: running: automake --add-missing --copy --force-missing
configure.ac:3: installing './install-sh'
configure.ac:3: installing './missing'
autoreconf: Leaving directory '.'
checking for a BSD-compatible install... /gnu/store/8awm79px539lwhg6gczh4h3nx5kfdj1c-profile/bin/install -c
checking whether build environment is sane... yes
checking for a race-free mkdir -p... /gnu/store/8awm79px539lwhg6gczh4h3nx5kfdj1c-profile/bin/mkdir -p
checking for gawk... gawk
checking whether make sets $(MAKE)... yes
checking whether make supports nested variables... yes
checking that generated files are newer than configure... done
configure: creating ./config.status
config.status: creating Makefile
make  dist-gzip am__post_remove_distdir='@:'
make[1]: Entering directory '/tmp/automake-repro'
make  distdir-am
make[2]: Entering directory '/tmp/automake-repro'
make[2]: *** No rule to make target 'ChangeLog', needed by 'distdir-am'.  Stop.
make[2]: Leaving directory '/tmp/automake-repro'
make[1]: *** [Makefile:261: distdir] Error 2
make[1]: Leaving directory '/tmp/automake-repro'
make: *** [Makefile:342: dist] Error 2
--8<---------------cut here---------------end--------------->8---

Notice that it tries to find ChangeLog, however fails to do so, since
the rule is commented out (due to the conditional).

After some more digging, I think the problem is handle_dist function,
specifically this part:

--8<---------------cut here---------------start------------->8---
      if (dir_has_case_matching_file ($relative_dir, $cfile)
	  # The file might be absent, but if it can be built it's ok.
	  || rule $cfile)
	{
	  push_dist_common ($cfile);
	}
--8<---------------cut here---------------end--------------->8---

There is this bullet point in the manual mentioning something like this:

> Automake has another built-in list of files automatically distributed
> if they are found either with the plain name, or with extension .md
> (presumably MarkDown, though this not checked). They are checked for
> in that order, so the plain name is preferred. These are: AUTHORS
> ChangeLog INSTALL NEWS README README-alpha THANKS.

So I think there are two problems:

1. The documentation does not mention this happens even if just a
   rule is found.
2. The code does not respect that the rule is if-ed away.

Have a nice day,
Tomas Volf

-- 
There are only two hard things in Computer Science:
cache invalidation, naming things and off-by-one errors.

Message #8 received at 79049 <at> debbugs.gnu.org (full text, mbox):

From: Karl Berry <karl <at> freefriends.org>
To: ~@wolfsden.cz
Cc: 79049 <at> debbugs.gnu.org
Subject: Re: bug#79049: ChangeLog is added to distribution even when inside a
 false conditional
Date: Sat, 19 Jul 2025 16:08:33 -0600

Thanks for the report and looking into the code. There have been
previous reports relating to targets and conditionals, e.g.,
https://debbugs.gnu.org/cgi/bugreport.cgi?bug=13940 (from 2013 ...)
and I think it's been mentioned in mail and elsewhere, too.

Updating the doc is no problem, thanks. But I'm not sure how to do a
real fix.  The conditional might be defined at configure-time, so
automake can't know whether it's true or not when it's running, as far
as I can see.

Maybe you or someone else has a clearer idea of how to proceed?
--thanks, karl.

Message #11 received at 79049 <at> debbugs.gnu.org (full text, mbox):

From: Tomas Volf <~@wolfsden.cz>
To: Karl Berry <karl <at> freefriends.org>
Cc: 79049 <at> debbugs.gnu.org
Subject: Re: bug#79049: ChangeLog is added to distribution even when inside
 a false conditional
Date: Sun, 20 Jul 2025 21:58:13 +0200

Karl Berry <karl <at> freefriends.org> writes:

> Thanks for the report and looking into the code. There have been
> previous reports relating to targets and conditionals, e.g.,
> https://debbugs.gnu.org/cgi/bugreport.cgi?bug=13940 (from 2013 ...)
> and I think it's been mentioned in mail and elsewhere, too.
>
> Updating the doc is no problem, thanks. But I'm not sure how to do a
> real fix.  The conditional might be defined at configure-time, so
> automake can't know whether it's true or not when it's running, as far
> as I can see.

Hm, this is very good point I failed to realize.

> Maybe you or someone else has a clearer idea of how to proceed?

Dropping the magic related to these special files is probably not
possible now, so only realistic option I see is to document the edge
case and move on.  Not great, so I hope someone will have a better
idea. :/

Tomas

-- 
There are only two hard things in Computer Science:
cache invalidation, naming things and off-by-one errors.

Previous Next

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