once again I did not have much time, so I kept the semi-automated tests going and fixed a handful of issues.
Python .pyc files became a larger problem since February with a new python3 release. I see new influences of ASLR into the bytecode. A comparable old bug on that topic is https://bugs.python.org/issue34033 Also we still declare many python packages as "noarch", even though .pyc files differ between architectures, so how can we really know that they are equivalent?
Java xmvn metadata is still a major headache with its random UUIDs https://bugzilla.opensuse.org/show_bug.cgi?id=1162112
I will host a workshop on reproducible builds this Friday https://events.opensuse.org/conferences/oSC22/program/proposals/3707
and will have a virtual presentation on https://ossna2022.sched.com/event/11NpJ/reproducible-builds-unexpected-benef...
Then there is the automated report: last month's status: https://email@example.com/thread/L...
Last months' reproducible builds project updates (including my work): https://reproducible-builds.org/reports/2022-04/
I uploaded https://rb.zq1.de/compare.factory-20220601/ today
https://rb.zq1.de/spec/glossar.txt explains the meaning of below values: total-packages: 14394 (+23) build-tried: 14386 (+24) build-failed: 44 (+16) build-n-a: 165 (-4) build-succeeded: 14177 (+11) build-official-failed+na: 216 (+143) build-compare-failed: 523 (+15) build-compare-succeeded: 13654 (-4) verify-failed: 429 (-48) verified-semi-reproducible: 12785 (+24) bit-by-bit-identical: 13411 (+9) not-bit-by-bit-identical: 781 (+9) not-bit-by-bit-identicalcheck: 766 (+2)
https://rb.zq1.de/compare.factory-20220601/graph.png shows the change over time
https://rb.zq1.de/compare.factory-20220601/unreproduciblerings.txt lists very unreproducible core packages (bootstrap+DVD)
Of the badly unreproducible packages, 4 were in ring0 66 were in ring1
That makes it 70/3316 => 2.11 % which is below the overall average of 523/14177 => 3.69 %
781/14177 => 5.51 % of packages are not perfectly reproducible
Ciao Bernhard M.