On Sat, Mar 14, 2020 at 10:15 PM Jim Meyering <j...@meyering.net> wrote: > When testing the latest automake with autoconf-2.69, all of automake's > tests pass. But with the latest autoconf (built from master) in my > path, two tests "FAIL", as below. I bisected to v2.69-181-g78ad1b0b, > aka > https://git.savannah.gnu.org/cgit/autoconf.git/commit/?id=v2.69-181-g78ad1b0b > > I plan to release automake-1.16.2 very soon (from master), maybe even > tomorrow. This does not block my planned automake release, but it'd be > great to resolve soon, nonetheless. > > FAIL: t/autohdr4 > ================ > > Running from installcheck: no > Test Protocol: none > PATH = /home/j/w/co/automake/bin:/home/j/w/co/automake/t/ax:/usr/share/Modul > es/bin:/usr/lib64/ccache:/home/j/bin:/p/bin:/usr/local/bin:/sbin:/usr/sbin:/ > bin:/usr/bin:/usr/games:/usr/local:/usr/local/bin > ++ pwd > /home/j/w/co/automake/t/autohdr4.dir > + cat > + mkdir sub1 sub2 sub3 > + : > + echo '#define NAME "grepme1"' > + cat > + cat > + cat > + aclocal-1.16 -Werror > + autoconf > + autoheader > autoheader: cannot open sub2/config.${BOT}: No such file or directory > + am_exit_trap 1 > + exit_status=1 > + set +e > + cd /home/j/w/co/automake > + test none = tap > + case $am_explicit_skips in > + test 1 -eq 0 > + keep_testdirs=yes > + am_keeping_testdirs > + case $keep_testdirs in > + return 0 > + set +x > FAIL t/autohdr4.sh (exit status: 1) > > FAIL: t/autohdr-subdir-pr12495 > ============================== > ... similar. > > Full disclosure, bisection details: > > # Build at each iteration with this: > autoreconf -vi && ./configure --prefix=/tmp/ac && make && ( > PATH=`pwd`/tests:$PATH; autoreconf -vi ) && make install > > # Test like this: > (cd ~/automake && PATH=/tmp/ac/bin:$PATH make check TESTS=t/autohdr4.sh)
Hi Daniel (Cc'd), since the problematic commit is from you, would you please see if you can adjust it not to cause that regression?
