postgresql-9.6.3-1.fc27 fails to build in F27 because hstore_plperlu test fails: for extra in contrib/hstore contrib/hstore_plperl; do make -C '../..'/$extra DESTDIR='/builddir/build/BUILD/postgresql-9.6.3'/tmp_install install >>'/builddir/build/BUILD/postgresql-9.6.3'/tmp_install/log/install.log || exit; done PATH="/builddir/build/BUILD/postgresql-9.6.3/tmp_install/usr/bin:$PATH" LD_LIBRARY_PATH="/builddir/build/BUILD/postgresql-9.6.3/tmp_install/usr/lib64" ../../src/test/regress/pg_regress --inputdir=. --temp-instance=./tmp_check --bindir= --dbname=contrib_regression hstore_plperl hstore_plperlu create_transform ============== creating temporary instance ============== ============== initializing database system ============== ============== starting postmaster ============== running on port 57835 with PID 42130 ============== creating database "contrib_regression" ============== CREATE DATABASE ALTER DATABASE ============== running regression test queries ============== test hstore_plperl ... ok test hstore_plperlu ... FAILED test create_transform ... ok ============== shutting down postmaster ============== ====================== 1 of 3 tests failed. ====================== [...] === trying to find all regression.diffs files in build directory === === make failure: contrib/hstore_plperl/regression.diffs === *** /builddir/build/BUILD/postgresql-9.6.3/contrib/hstore_plperl/expected/hstore_plperlu.out Mon May 8 21:15:12 2017 --- /builddir/build/BUILD/postgresql-9.6.3/contrib/hstore_plperl/results/hstore_plperlu.out Wed May 17 20:01:44 2017 *************** *** 150,158 **** INFO: $VAR1 = { 'a' => '1', 'b' => { ! 'aa' => 'bb', ! 'cc' => undef ! } }; SELECT * FROM test1; --- 150,158 ---- INFO: $VAR1 = { 'a' => '1', 'b' => { ! 'aa' => 'bb', ! 'cc' => undef ! } }; SELECT * FROM test1; ====================================================================== A difference between working and failing build root is: python2-devel 2.7.13-8.fc27 > 2.7.13-9.fc27 systemtap-sdt-devel 3.2-0.20170410git... > 3.2-0.20170512git... python3-devel 3.6.1-6.fc27 > 3.6.1-7.fc27 util-linux 2.29.1-2.fc26 > 2.30-0.1.fc27 perl-Data-Dumper 2.161-2.fc26 > 2.167-1.fc27 python3 3.6.1-6.fc27 > 3.6.1-7.fc27 glib2 2.52.2-1.fc27 > 2.52.2-2.fc27 system-python-libs 3.6.1-6.fc27 > 3.6.1-7.fc27 libmount 2.29.1-2.fc26 > 2.30-0.1.fc27 libuuid 2.29.1-2.fc26 > 2.30-0.1.fc27 python2 2.7.13-8.fc27 > 2.7.13-9.fc27 python3-libs 3.6.1-6.fc27 > 3.6.1-7.fc27 python3-rpm-generators > 4.13.0.1-1.fc27 python2-libs 2.7.13-8.fc27 > 2.7.13-9.fc27 libacl 2.2.52-13.fc26 > 2.2.52-14.fc27 libsmartcols 2.29.1-2.fc26 > 2.30-0.1.fc27 libfdisk 2.29.1-2.fc26 > 2.30-0.1.fc27 libblkid 2.29.1-2.fc26 > 2.30-0.1.fc27 kernel-headers 4.12.0-0.rc0.git7.1.... > 4.12.0-0.rc1.git1.1.... coreutils 8.27-8.fc27 > 8.27-9.fc27 device-mapper-libs 1.02.140-1.fc27 > 1.02.140-2.fc27 perl-Term-ANSIColor 4.06-2.fc26 > 4.06-3.fc27 perl-Unicode-Normalize 1.25-366.fc26 > 1.25-367.fc27 curl 7.54.0-4.fc27 > 7.54.0-5.fc27 device-mapper 1.02.140-1.fc27 > 1.02.140-2.fc27 coreutils-common 8.27-8.fc27 > 8.27-9.fc27 gnupg2 2.1.20-2.fc27 > 2.1.21-2.fc27 system-python 3.6.1-6.fc27 > 3.6.1-7.fc27 acl 2.2.52-13.fc26 > 2.2.52-14.fc27 libcurl 7.54.0-4.fc27 > 7.54.0-5.fc27 gnutls 3.5.12-1.fc27 > 3.5.12-2.fc27 libusbx > 1.0.21-2.fc26 I suspect this could be caused by upgrading perl-Data-Dumper from 2.161-2.fc26 to 2.167-1.fc27 (different output indentation). The Data-Dumper-2.167 will be released by upstream within perl-5.26.0 in few weeks. Upstream can uses Perl 5.26.0 RC1 for testing now.
Thanks for the report. Upstream is already aware of this: https://www.postgresql.org/message-id/24215.1494714163%40sss.pgh.pa.us They are aiming to fix this by using $Data::Dumper::Indent set to 0
Actually the fix is already pushed upstream so I have added the fix to rawhide: http://pkgs.fedoraproject.org/cgit/rpms/postgresql.git/commit/?id=5b2a26a63d4f706634bdf66ffa97ae5f77bc941b
looks like this issue is back (not sure if exact the same but the same test failed in rawhide) https://koji.fedoraproject.org/koji/taskinfo?taskID=20026948 https://kojipkgs.fedoraproject.org//work/tasks/6949/20026949/build.log
(In reply to Filipe Rosset from comment #3) > looks like this issue is back (not sure if exact the same but the same test > failed in rawhide) > > https://koji.fedoraproject.org/koji/taskinfo?taskID=20026948 > https://kojipkgs.fedoraproject.org//work/tasks/6949/20026949/build.log nevermind, it's fixed in 9.6.3-2, sorry for the noise.