Reproducible Builds: Weekly report #72

What happened in the Reproducible Builds effort between Sunday September 4 and Saturday September 10 2016:

Reproducible work in other projects

Python 3.6’s dictonary type now retains the insertion order. Thanks to themill for the report.

In coreboot, Alexander Couzens committed a change to make their release archives reproducible.

Patches submitted

• #836609 filed against nostalgy by Chris Lamb.
• #836605 filed against torque by Chris Lamb.
• #836968 filed against erlang-p1-oauth2 by Chris Lamb.
• #836970 filed against erlang-p1-sqlite3 by Chris Lamb.
• #836817 filed against tj3 by Chris Lamb.
• #836969 filed against erlang-p1-xmlrpc by Chris Lamb.

Reviews of unreproducible packages

We’ve been adding to our knowledge about identified issues. 3 issue types have been added:

• random_id_in_pdf_generated_by_dblatex
• captures_execution_time
• captures_home_dir

1 issue type has been updated:

• Expand comment for random_id_in_pdf_generated_by_dblatex

16 have been have updated:

• Add patch for nostalgy
• Add patch for torque
• Expand comment for babl.
• Tag pbuilder with random_id_in_pdf_generated_by_dblatex
• Tag uclibc with users_and_groups_in_tarball
• Tag realmd with different_encoding_in_html_by_docbook_xsl
• Tag mrmpi with timestamps_in_documentation_generated_by_htmldoc
• Tag dynare with leaks_path_environment_variable
• Tag sdlgfx with timestamps_in_tarball
• Tag pantomime1.2 with plist_weirdness
• Tag blockattack with different_due_to_umask
• Tag expeyes with different_pot_creation_date_in_gettext_mo_files
• Tag android-platform-frameworks-data-binding with random_ordering_in_pom
• Tag krunner with users_and_groups_in_tarball
• Tag libevocosm with leaks_path_environment_variable
• Tag gdbm with random_order_in_md5sums

13 have been removed, not including removed packages:

• htp fixed in/since 1.19-2
• Remove openni-sensor-pointclouds and openni-sensor-primesense; fixed by __DATE__ & __TIME__
• hsqldb fixed in/since 2.3.3+dfsg2-1
• linux-tools RM
• ksnapshot RM
• cobalt-panel-utils RM
• strigi RM
• libkdeedu RM
• erc RM
• ttf-atarismall RM
• gnupg-doc RM
• remove RMed django-localflavor
• easymp3gain RM

100s of packages have been tagged with the more generic captures_build_path, and many with captures_kernel_version, user_hostname_manually_added_requiring_further_investigation, user_hostname_manually_added_requiring_further_investigation, captures_shell_variable_in_autofoo_script, etc.

Particular thanks to Emanuel Bronshtein for his work here.

Weekly QA work

FTBFS bugs have been reported by:

• Aaron M. Ucko (1)
• Chris Lamb (7)

diffoscope development

• Mattia Rizzolo:
  • Force LC_ALL=C.UTF-8 in basic-command-list autopkgtest so that diffoscope can always output something
• Ximin Luo:
  • html-dir: use 4 * the given max-diff-block-lines
  • difference: Don’t truncate diff output by default, and use a separate flag when doing so
  • html-dir: In html-dir output, split large diffs across several child pages
  • Rename no-max-limits to no-default-limits, a more accurate name
  • Increase default max_diff_input_lines, should be OK based on what we’ve seen
  • Fix behaviour for max-diff-block-limit=1 and don’t report when 0 lines are removed
  • Set limits to inf when parsing args, instead of checking for “> 0” everywhere
  • Suppress decimal line numbers for xxd output

strip-nondeterminism development

• Chris Lamb:
  • handlers/jar.pm: Reintroduce subroutine prototype on _jar_filename_cmp to fix ordering regression in .jar files.
  • bin/strip-nondeterminism: die if file does not exist (Closes: #800159)
  • t/fixtures/jar/*: Add a testcase for jar file ordering and normalisation.
  • t/fixtures.t:
    • Check that output matches expected before comparing attributes to make it less confusing on failure.
    • Don’t test whether inode number has changed either.
    • Add a lookup table for descriptions of stat(1) indices.
    • Re-order @fixtures setup for global->specific ordering.
    • Re-add top-level “plan” call dropped in previous commit.
    • Align size check with atime, etc.
    • Make it clearer that we are checking for attributes
    • Add comment why we use a per-fixture temporary directory.
  • Drop all calls to unlink_on_destroy now that we copy the file. (Closes: #836670)
  • Extend testsuite to ensure no extra files are leftover again.

tests.reproducible-builds.org:

• F-Droid:
  • Hans-Christoph Steiner found after extensive debugging that for kvm-on-kvm, vagrant from stretch is needed (or a backport, but that seems harder than setting up a new VM).
• FreeBSD:
  • Holger updated the VM for testing FreeBSD to FreeBSD 10.3.

Misc.

This week’s edition was written by Chris Lamb and Holger Levsen and reviewed by a bunch of Reproducible Builds folks on IRC.