1. 18 Mar, 2021 4 commits
    • Michael Pyne's avatar
      Address undefined behavior during shutdown. · fce79085
      Michael Pyne authored
      Event filters strike again, along with our old friend, sibling ordering
      dependencies not captured by the Qt memory hierarchy. Again, noted by
      ubsan.
      fce79085
    • Michael Pyne's avatar
      playlist: Avoid resizing columns until after we've generated weights. · 66ddbb3b
      Michael Pyne authored
      This avoids warnings about undefined use of NaN and similar, when we try
      to resize the columns too early (before there are any tracks to use to
      inform average column widths).
      66ddbb3b
    • Michael Pyne's avatar
      Fix undefined behavior in PlaylistItem. · b67785f9
      Michael Pyne authored
      The PlaylistItem ctor meant to be called from its subclass's constructor
      (CollectionListItem) copied the this pointer too early. When
      PlaylistItem is being constructed, by C++ rules the PlaylistItem is not
      a CollectionListItem yet. This was appropriately flagged by ubsan.
      
      I fixed this by ensuring the only user of this subclass fixes up the
      pointer as soon as it can, and making it a private constructor so it's
      not used elsewhere by mistake.
      b67785f9
    • Michael Pyne's avatar
      Fix use of undefined behavior in status label. · 10110d76
      Michael Pyne authored
      The event filter doesn't check that the received event is a QMouseEvent
      but it is possible (and actually happened) to receive other events from
      Qt.
      
      But in this case we don't even need an event filter, we're just handling
      mouse events and Qt provides entirely usable methods for that already.
      10110d76
  2. 13 Mar, 2021 1 commit
  3. 24 Feb, 2021 1 commit
  4. 23 Feb, 2021 1 commit
    • Michael Pyne's avatar
      clazy: Fix most remaining warnings. · 288c69eb
      Michael Pyne authored
      I'm not going to worry about non-POD statics in an application, and I
      can't fix the Q_PROPERTY warnings in Phonon. Other than that, all
      default clazy warnings should be gone now.
      
      Of note is the tagtransactionmanager.cpp setFile call, which is not just
      a warning but an actual bug (good job clazy!).
      288c69eb
  5. 22 Feb, 2021 5 commits
  6. 13 Feb, 2021 2 commits
    • Michael Pyne's avatar
      test: Fix tagguessertest lookup error with ctest. · 3a3bc5e5
      Michael Pyne authored
      For whatever reason, running plain ctest or even "ninja test" on a
      testing-enabled does not succeed in finding tagguessertest to run it,
      probably because this test is located in the build directory's /bin
      subfolder and the test metadata is incorrect.
      
      However I was able to solve this just by using the existing ECM macro to
      add a single test (https://api.kde.org/ecm/module/ECMAddTests.html)
      directly, so that's what I did. ctest now runs and passes both tests
      fine again.
      3a3bc5e5
    • Michael Pyne's avatar
      cmake: Remove needless checks for Taglib format support. · 082914b0
      Michael Pyne authored
      The recent commit 9a54d6e7 to use
      FindTaglib from ECM breaks detection of Ogg Opus support in Taglib for
      some reason. This causes crashes when playing Ogg Opus files because
      their relevant Taglib::File subclass is not created from our MediaFiles
      factory.
      
      However the check is needless. Taglib support for ASF and MP4 files has
      been standard since 2012 or so, and Ogg Opus support was made standard
      around the same time frame. So rather than check for these as optional
      features, just assume that supported versions of Taglib support these
      formats.
      082914b0
  7. 11 Feb, 2021 1 commit
    • Michael Pyne's avatar
      Merge branch 'neoninteger/juk-fix-rename-folders' · 8956f56d
      Michael Pyne authored
      Merging a change from Callum Parsey to fix a bug with the file renamer
      where it was, in essence, trying to treat the resulting file basename as
      if it were another directory in the path, and failing as a result.
      
      This comes extremely late, as I somehow missed this merge request
      hanging out for JuK until after an unrelated one came in recently. But
      no excuses, that's my fault. My apologies to the submitter.
      
      I didn't see a corresponding bug to close. This fix should land in
      version 20.12.3 and later.
      8956f56d
  8. 10 Feb, 2021 1 commit
  9. 06 Feb, 2021 3 commits
  10. 29 Jan, 2021 1 commit
  11. 02 Jan, 2021 1 commit
  12. 01 Jan, 2021 1 commit
  13. 02 Dec, 2020 1 commit
  14. 28 Nov, 2020 1 commit
  15. 08 Nov, 2020 1 commit
  16. 31 Oct, 2020 1 commit
  17. 03 Oct, 2020 1 commit
  18. 10 Sep, 2020 1 commit
  19. 31 Aug, 2020 2 commits
  20. 05 Aug, 2020 1 commit
  21. 28 Jul, 2020 1 commit
  22. 13 Jul, 2020 1 commit
  23. 12 Jul, 2020 1 commit
  24. 11 Jul, 2020 2 commits
  25. 03 Jul, 2020 1 commit
  26. 19 Jun, 2020 3 commits
    • Luca Beltrame's avatar
      Merge branch 'release/20.04' · 0b6a6022
      Luca Beltrame authored
      0b6a6022
    • Luca Beltrame's avatar
      Fix CMake module search · 2079a775
      Luca Beltrame authored
      The local cmake modules should, in this case, be preferred compared to
      the ones in ECM, or the build will not be probably configured,
      causing includes not to be found, and thus ending up in failure.
      2079a775
    • Luca Beltrame's avatar
      Fix CMake module search · 3303552d
      Luca Beltrame authored
      The local cmake modules should, in this case, be preferred compared to
      the ones in ECM, or the build will not be probably configured,
      causing includes not to be found, and thus ending up in failure.
      3303552d