Re: [Platform-testers] Automake 1.11.1b test release

[Stefano Lattarini]

Reference:
 <http://lists.gnu.org/archive/html/automake/2011-12/msg00039.html>

[Adding automake-patches and bug-automake]
[Answers will drop the automake list]

-*-*-

Hi Bruno; thanks for the testing and the quick feedback.

On Friday 16 December 2011, Bruno Haible wrote:
> > The pre-release automake version 1.11.1b is now available at
> > <ftp://alpha.gnu.org/gnu/automake>.
> 
> On machines without a sufficiently new 'perl', configure fails:
> 
>   checking for perl... no
>   configure: error: perl not found
> 
>   checking for perl... /bin/perl
>   Perl 5.006 required--this is only version 5.00503, stopped at -e line 1.
>   configure: error: perl 5.6 or better is required; perl 5.8.2 or better
>   is recommended.  If you have several perl versions
>   installed, select the one Automake should use using
>     ./configure PERL=/path/to/perl
>
"git blame" tells me that the perl 5.6 requirement has been introduced
*seven* years ago by Alexandre Duret-Lutz, in commit 8735b00e.  Since
nobody has complained about that until now, I assume the requirement is
reasonable for real-world usages.  Are you sure the machines you've
tested on don't have a more modern version of perl installed elsewhere?

>   checking for perl... /usr/sbin/perl
>   Perl 5.006 required--this is only version 5.00405, stopped at -e line 1.
>   configure: error: perl 5.6 or better is required; perl 5.8.2 or better
>   is recommended.  If you have several perl versions
>   installed, select the one Automake should use using
>     ./configure PERL=/path/to/perl
> 
> On machines without a sufficiently new 'autoconf', configure fails:
> 
>   checking whether autoconf is installed... no
>   configure: error: Autoconf 2.62 or better is required.
>       Please make sure it is installed and in your PATH.
>
Ouch.  Still, I have no plan to relax the requirement of the autoconf
minimal version, again for the reason that no one has complained about
that requirement (in place since automake 1.11) so far, so that we can
assume it is reasonable for real-world usages.  If you have the time
and the inclination, you could install a more modern version of autoconf
on those older systems, and try again; consider that I still test
automake with autoconf 2.62 quite regularly, so the 2.62 version should
truly be enough (so that hopefully you won't alos need to install a
newer version og GNU m4, as required by the latest autoconf versions).

> On Linux/x86_64:
> 
>   All 823 tests behaved as expected (11 expected failures)
>
Good.

> On MacOS X 10.5:
> 
>   All 812 tests behaved as expected (11 expected failures)
>
Wow, and I've never even tested on MacOS X :-)

> On NetBSD 5.1:
> 
>   All 760 tests behaved as expected (10 expected failures)
>
JFTR, have the libtool tests run there?  Anyway, nice to see a clean
testsuite run on a BSD system.

> On Linux/IA-64:
> 
>   FAIL: acloca20.test
>   FAIL: aclocal-path-install-serial.test
>   FAIL: autodist-aclocal-m4.test
>   FAIL: help-python.test
>   FAIL: missing5.test
>   FAIL: obsolete.test
>   FAIL: tagsub.test
>   7 of 824 tests failed
> 
>   Find attached the test suite log file. Apparently some of the failures are
>   due to the autoconf version:
> 
>   $ autoconf --version
>   Autoconf version 2.13
>   ---
>   Autoconf 2.13 chosen by Debian wrapper script.
>   For information and tuning advice see autoconf(1).
>
Duh, the infamous autoconf wrapper script :-(

The first attached patch should solve this issue for the moment (tested
on Debian unstable).  In the long run, I think we shoud switch from
`configure.in' to `configure.ac' in all our test cases, at which point
the wrapper should do the right thing automatically.

> On FreeBSD 6.4:
> 
>   FAIL: man4.test
>   FAIL: yacc-dist-nobuild-subdir.test
>
Known and long-standing failures, no regression here.  See also:
 <http://debbugs.gnu.org/cgi/bugreport.cgi?bug=7884>

>   XPASS: remake-m4-pr10111.test
>
Oh, funny, the same FreeBSD bug thay is causing automake bug#7884 and
the failure in yacc-dist-nobuild-subdir.test is masking the automake
limitation that should cause this test to fail.  Not worth worrying
about FTM I think (but it would be nice to solve the issue for the
future 1.11.3 maintenance release).

BTW, `remake-am-pr10111.test' should have XPASSed for the same reasons
`remake-m4-pr10111.test' does, but this is not the case; that was due
to another, unrelated bug in the test case, that is fixed by the second
attached patch.

>   3 of 766 tests did not behave as expected (1 unexpected pass)
>
>   Find attached the test suite log file. Probably the cause is that
>   on this machine, /usr/bin/make is being used, which is BSD make,
>   not GNU make.
>
Exactly.  But that should be the case also for the NetBSD system you
have tested on, right?

> Bruno
> 

Thanks,
  Stafano

From e1fbb22d951f65db7015d9e31fe7713df75e1384 Mon Sep 17 00:00:00 2001
Message-Id: <address@hidden>
From: Stefano Lattarini <address@hidden>
Date: Fri, 16 Dec 2011 11:12:31 +0100
Subject: [PATCH] test defs: hack to support autoconf-wrapper programs

* tests/defs.in ($AUTOCONF): Add a dummy `-B' option to the
autoconf invocation, so that, when the Debian autoconf wrapper
is involved, it will correctly dispatch an autoconf >= 2.50
instead of defaulting to autoconf 2.13.
($AUTOHEADER, $AUTORECONF): Likewise, but for autoheader and
autoreconf respectively.

Reported by Bruno Haible:
<http://lists.gnu.org/archive/html/automake/2011-12/msg00039.html>
---
 ChangeLog     |   12 ++++++++++++
 tests/defs.in |   11 +++++++++++
 2 files changed, 23 insertions(+), 0 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index 6bc7daa..9f9b3cc 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,17 @@
 2011-12-16  Stefano Lattarini  <address@hidden>
 
+       test defs: hack to support autoconf-wrapper programs
+       * tests/defs.in ($AUTOCONF): Add a dummy `-B' option to the
+       autoconf invocation, so that, when the Debian autoconf wrapper
+       is involved, it will correctly dispatch an autoconf >= 2.50
+       instead of defaulting to autoconf 2.13.
+       ($AUTOHEADER, $AUTORECONF): Likewise, but for autoheader and
+       autoreconf respectively.
+       Reported by Bruno Haible:
+       <http://lists.gnu.org/archive/html/automake/2011-12/msg00039.html>
+
+2011-12-16  Stefano Lattarini  <address@hidden>
+
        tests: fix a minor spurious failure with FreeBSD make
        * tests/remake-am-pr10111.test: Avoid using `#' comments in
        makefile recipes, as these have been confusing FreeBSD make.
diff --git a/tests/defs.in b/tests/defs.in
index f24d2ad..104e60e 100644
--- a/tests/defs.in
+++ b/tests/defs.in
@@ -103,6 +103,17 @@ test -z "$AUTORECONF" && AUTORECONF="@am_AUTORECONF@"
 test -z "$AUTOHEADER" && AUTOHEADER="@am_AUTOHEADER@"
 test -z "$AUTOUPDATE" && AUTOUPDATE="@am_AUTOUPDATE@"
 test -z "$MISSING" && MISSING=`pwd`/../lib/missing
+
+# This is an hack to seamlessly support the infamous "autoconf wrappers",
+# that might dispatch different autoconf versions depending on the name
+# of the input files and/or the command-line options used.  See:
+# <http://lists.gnu.org/archive/html/automake/2011-12/msg00039.html>
+# FIXME: in the long run, the better fix will be to convert our testsuite
+# to use `configure.ac' instead of `configure.in' as autoconf input.
+AUTOCONF="$AUTOCONF -B /no/such/dir"
+AUTOM4TE="$AUTOM4TE -B /no/such/dir"
+AUTORECONF="$AUTORECONF -B /no/such/dir"
+
 # Use -Werror because this also turns some Perl warnings into error.
 # (Tests for which this is inappropriate should use -Wno-error.)
 test -z "$ACLOCAL" && ACLOCAL="aclocal-$APIVERSION -Werror"
-- 
1.7.2.3

From 0141e1fb6502985e7e3483106f918136ff53011c Mon Sep 17 00:00:00 2001
Message-Id: <address@hidden>
From: Stefano Lattarini <address@hidden>
Date: Fri, 16 Dec 2011 01:46:14 +0100
Subject: [PATCH] tests: fix a minor spurious failure with FreeBSD make

* tests/remake-am-pr10111.test: Avoid using `#' comments in
makefile recipes, as these have been confusing FreeBSD make.
The failure was masked by the fact that this test is currently
expected to fail.

Suggested by a report from Bruno Haible.
---
 ChangeLog                    |    9 +++++++++
 tests/remake-am-pr10111.test |    7 ++++---
 2 files changed, 13 insertions(+), 3 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index f61872d..6bc7daa 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,14 @@
 2011-12-16  Stefano Lattarini  <address@hidden>
 
+       tests: fix a minor spurious failure with FreeBSD make
+       * tests/remake-am-pr10111.test: Avoid using `#' comments in
+       makefile recipes, as these have been confusing FreeBSD make.
+       The failure was masked by the fact that this test is currently
+       expected to fail.
+       Suggested by a report from Bruno Haible.
+
+2011-12-16  Stefano Lattarini  <address@hidden>
+
        tests: make two test scripts executable
        * tests/remake-am-pr10111.test: Make executable.
        * tests/remake-m4-pr10111.test: Likewise.
diff --git a/tests/remake-am-pr10111.test b/tests/remake-am-pr10111.test
index 6f622f8..f7ad771 100755
--- a/tests/remake-am-pr10111.test
+++ b/tests/remake-am-pr10111.test
@@ -35,9 +35,10 @@ END
 cat > Makefile.am <<'END'
 include $(srcdir)/foobar.am
 $(srcdir)/foobar.am:
-## Creative quoting to avoid spurious matches in the grepping
-## of Makefile.in, later.
-       echo '#' 'foobar' 'was 'here' '#' > $@
+## Use of escape sequences with printf instead of literal `#' characters
+## is to avoid issues with FreeBSD make.  Creative quoting and spacing is
+## to avoid spurious matches in the grepping of Makefile.in, later.
+       printf '\043 %s %s \043\n' foobar "was here" > $@
 END
 
 $ACLOCAL
-- 
1.7.2.3