Hi, last month's status: https://lists.opensuse.org/opensuse-factory/2020-04/msg00026.html Last months' reproducible builds project updates (including my work): https://reproducible-builds.org/reports/2020-03/ I uploaded https://rb.zq1.de/compare.factory-20200430/ today and rbstats are: total-packages: 12788 (+7) build-tried: 12782 (+10) build-failed: 48 (-6) build-n-a: 148 (+62) build-succeeded: 12586 (-46) build-official-failed+na: 237 (+102) build-compare-failed: 462 (+54) build-compare-succeeded: 12124 (-100) verify-failed: 516 (-10) verified-semi-reproducible: 11342 (-116) verified-bit-identical: 0 (+0) bit-by-bit-identical: 11796 (-66) not-bit-by-bit-identical: 790 (+26) not-bit-by-bit-identicalcheck: 790 (+20) https://rb.zq1.de/compare.factory-20200430/graph.png shows the change over time https://rb.zq1.de/compare.factory-20200430/unreproduciblerings.txt lists very unreproducible core packages (bootstrap+DVD) Of the badly unreproducible packages, 3 were in ring0 51 were in ring1 That makes it 54/2901 => 1.86 % which is below the overall average of 462/12586 => 3.67 % 790/12586 => 6.28 % of packages are not perfectly reproducible newly unreproducible core packages: glassfish-servlet-api java gtk-sharp2 monodoc uses nondeterministic filesystem readdir order libqt5-qtscript filesys readdir order from Qt podman random go tmp build path python-debtcollector https://review.openstack.org/652669 sphinx .pickle file marshalling python-numpy .pyc file marshalling tigervnc has signature with random RSA key xerces-j2 java yudit ASLR? uninitialized bytes in .my files created by "mytool" packages that are problematic to build in my setup, but not OBS: breezy -j1 diy many tests fail frr -j1 lua-toluapp ?scons fail perl-Image-Sane https://bugzilla.opensuse.org/show_bug.cgi?id=1170639 stuck build on -j1 python2-pip ImportError: No module named six python3-espressomd -j1 failed all_compare_test python-astropy https://github.com/astropy/astropy/issues/10228 python-keystonemiddleware expired SSL cert python-M2Crypto https://gitlab.com/m2crypto/m2crypto/-/issues/275 python-nbclient test_run_all_notebooks failed python-pyenchant test_docstrings failed python-ruffus test_with_logger.py Fatal Python error: Aborted tdiff https://github.com/F-i-f/tdiff/issues/2 vim https://bugzilla.opensuse.org/show_bug.cgi?id=1157623 Ciao Bernhard M. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Am 30.04.20 um 09:48 schrieb Jan Engelhardt:

On Thursday 2020-04-30 08:26, Bernhard M. Wiedemann wrote:

...

https://rb.zq1.de/compare.factory-20200430/colm.txt

colm is a case of ASLR.

The data it outputs wants to have some unique identifiers among all the identifiers it outputs, and it simply uses the pointer address of those objects while serializing to text.

Do you already have a "standard solution" (abstract way to address serialization of pointers) for cases like these?

String name( 128, "_ign_%p", regionSet->tokenIgnore );

If not, I'll be trying to use a global counter or something.

Yes, global counters is one way - as long as everything happens in one process, otherwise racy parallel processing can introduce nondeterminism. A workaround is to use setarch -R to disable ASLR during build. Otherwise sorting by something that is unique and deterministic also does the trick. E.g. these pointers are pointing to some data? strcmp/memcmp could tell you which comes first... -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org