1. 14 Apr, 2020 1 commit
    • Harald Sitter's avatar
      adjust smb kcm size to be sufficient vis a vis headers · f2c5e10d
      Harald Sitter authored
      Summary:
      we use Stretch to get a somewhat uniform appearance between the two views
      but Stretch is bit meh when it comes to actual sizing as it acts kind of
      like QSizePolicy::Preferred in that it allows shrinking columns so small
      that even the header gets cut off. this of course looks silly, so we
      now set the largest required width as minimal section width. this prevents
      the user from cutting off headers by resizing the window.
      
      secondly, and entirely related to that, because KCMs are embedded into
      a scrollview their sizehint kinda doesn't make any difference for
      the window size which then combined with the aforementioned stretch problem
      can lead to cut off headers by default as the window size is too small
      making the stretch cut into header text.
      to ensure the default size is reasonable let the view adjust to content
      on first show, effectively making the window as large as needed
      regardless of implicit hints.
      
      BUG: 419786
      FIXED-IN: 5.18.5
      
      Test Plan:
      - empty dialog
      - dialog with only shares
      - dialog with shares and mounts
      
      Reviewers: ngraham
      
      Reviewed By: ngraham
      
      Subscribers: plasma-devel
      
      Tags: #plasma
      
      Differential Revision: https://phabricator.kde.org/D28678
      f2c5e10d
  2. 03 Apr, 2020 2 commits
    • Harald Sitter's avatar
      strip non printable characters from opengl dri info · a20ea703
      Harald Sitter authored
      Summary:
      previously this was incorrectly indented
      
      was
      | Vendor | \s\s\s\s\s\s\s\sFoobar |
      
      is now
      | Vendor | Foobar                 |
      
      Test Plan: with D27980 applied the kernel module doesn't stick out anymore as now all strings are uniformly simplified and have no unexpected indentation
      
      Reviewers: apol
      
      Reviewed By: apol
      
      Subscribers: plasma-devel
      
      Tags: #plasma
      
      Differential Revision: https://phabricator.kde.org/D27981
      a20ea703
    • Harald Sitter's avatar
      make opengl module support multiple dri devices · 72a09a3f
      Harald Sitter authored
      Summary:
      previously the code would only ever look at dri0 when that may not actually
      be the device that is used. instead list all possible devices (I am
      assuming in nvidia/intel combos the actual computations may happen on any
      of them depending on configuration).
      as a side effect we can no longer list a single kernel module but instead
      list the module per-device.
      the devices use cardN names on linux.
      for the /proc based fallback we do not use cardN but simply N (to be
      improved in master with localized string).
      
      the legacy /proc/dri/ support was entirely removed. it's not a thing since
      linux 3.12, which is from 2013.
      
      on freebsd we continue to only support one device, in part because I have
      no way to test multi-device support. the info is woefully underwhelming and
      could as well not be there anyway TBH.
      
      BUG: 417986
      FIXED-IN: 5.18.5
      
      Test Plan: vm with 3 cards show all of them with suitable info
      
      Reviewers: apol
      
      Reviewed By: apol
      
      Subscribers: pino, zzag, apol, adridg, plasma-devel
      
      Tags: #plasma
      
      Differential Revision: https://phabricator.kde.org/D27980
      72a09a3f
  3. 31 Mar, 2020 1 commit
  4. 25 Mar, 2020 2 commits
    • Harald Sitter's avatar
      only add devices to the smb mount model that are network shares · 23d52fa3
      Harald Sitter authored
      Summary:
      previously we'd add all devices which would then cause a crash when
      trying to access them ->as<NetworkShare>() because they aren't
      
      Test Plan: have kcm open, plug in usb stick, observe no crash
      
      Reviewers: ngraham, broulik
      
      Reviewed By: ngraham, broulik
      
      Subscribers: broulik, plasma-devel
      
      Tags: #plasma
      
      Differential Revision: https://phabricator.kde.org/D28276
      23d52fa3
    • Harald Sitter's avatar
      skip over supposed network shares that aren't · 4847c9e2
      Harald Sitter authored
      Summary:
      we explicitly request listFromType(Solid::DeviceInterface::NetworkShare)
      but apparently still can get devices that aren't actually of that type.
      add an explicit type check to guard against bogus shares coming out of
      listFromType.
      
      CCBUG: 419220
      
      Test Plan: mount an overlay (I am guessing that's what the problem with kubuntu is), observe no crash on init
      
      Reviewers: ngraham
      
      Reviewed By: ngraham
      
      Subscribers: plasma-devel
      
      Tags: #plasma
      
      Differential Revision: https://phabricator.kde.org/D28275
      4847c9e2
  5. 16 Mar, 2020 1 commit
  6. 10 Mar, 2020 1 commit
  7. 25 Feb, 2020 1 commit
  8. 18 Feb, 2020 2 commits
    • Jonathan Esk-Riddell's avatar
      Update version number for 5.18.1 · 27380c8d
      Jonathan Esk-Riddell authored
      GIT_SILENT
      27380c8d
    • Harald Sitter's avatar
      make sure ifa_addr isn't null · 8cb74953
      Harald Sitter authored
      Summary:
      the manpage:
             The ifa_addr field points to a structure containing the interface
             address.  (The sa_family subfield should be consulted to determine
             the format of the address structure.)  This field may contain a null
             pointer.
      
      why that exactly happens I do not know in absence of more debug info.
      I expect it's something to do with an interface being up but not quite
      somehow. I utterly failed to simulate this though :|
      because of that I do not know if such a device has any use being listed,
      and I am adding some possibly relevant debug output in the hopes of
      maybe getting some data reported.
      if these devices still turn out useful, the MyNIC construction needs
      partially moving before the ifa_addr check.
      
      BUG: 417798
      FIXED-IN: 5.18.1
      
      Test Plan: hoping for the best!
      
      Reviewers: ngraham
      
      Subscribers: plasma-devel
      
      Tags: #plasma
      
      Differential Revision: https://phabricator.kde.org/D27467
      8cb74953
  9. 16 Feb, 2020 2 commits
  10. 08 Feb, 2020 2 commits
    • Harald Sitter's avatar
      import currently used license copies for reuse compliance · 4ea9df96
      Harald Sitter authored
      in accordance with our revised licensing policy.
      the previous files COPYING, .DOC and .LIB are respectively replaced by
      GPL-2.0-only, GFDL-1.2-only, LGPL-2.1-only
      4ea9df96
    • Harald Sitter's avatar
      replace samba module with data that works · f1768f7b
      Harald Sitter authored
      Summary:
      ... and doesn't require lots of maintenance!
      this targets 5.18 but is kind of unfortunate because it needs a bunch of
      new strings as no existing strings provide what is needed here.
      
      the previous module was super broken in various ways.
      in the interest of maintainability I've thrown everything out and replaced
      it with 2 core features which require only modeling code on the KInfoCenter
      end and provide actually (possibly) useful functionality to the design
      personas of plasma.
      
      there is now a single page which contains two table views:
      
      a) Exports: this is a simple table of shares "exported" from the host. the
         data for this comes from KSambaShare in KIOCore. this is the same API
         used by dolphin to create new shares, so the data will be consistent
         and the API needs maintaining anyway
      b) Imports: simple table of shares "imported" (i.e. mounted) onto the host.
         the data for this comes from solid
      
      both are backed by models, with an eye towards a future port to qml (out of
      scope since I want this fixed for 5.18)
      
      all previous functionality was removed, in part because it was doing CLI
      parsing, some of the parsing was broken, some of the CLI tools couldn't
      even run as !root, log parsing could use incorrect paths on existing users,
      log parsing has nothing to parse with samba defaults, log parsing didn't
      implement per-host log file support (current default in samba), log parsing
      didn't correctly implement per-user-config-log-file support.
      in short: there was not a single feature that worked properly.
      
      BUG: 411433
      BUG: 374141
      BUG: 325951
      FIXED-IN: 5.18.1
      
      {F8053130}
      
      Test Plan:
      # exports
      
      - nothing when nothing is exported
      - changing exports via dolphin is immediately reflected in the kcm
      - export data is valid
      
      # imports
      
      - nothing when nothing is mounted
      - (un)mounting a cifs updates the view immediately
      - data is valid
      
      Reviewers: #localization, #plasma, #vdg, davidedmundson
      
      Reviewed By: #plasma, #vdg, davidedmundson
      
      Subscribers: jriddell, davidedmundson, ltoscano, yurchor, ngraham, alexde, plasma-devel
      
      Tags: #plasma
      
      Differential Revision: https://phabricator.kde.org/D27061
      f1768f7b
  11. 06 Feb, 2020 1 commit
  12. 31 Jan, 2020 2 commits
    • Harald Sitter's avatar
      fix var name in sa_len branch · d7ba51de
      Harald Sitter authored
      s/buf/hostOut
      
      also line break the two conditional branches the same way so it's easier
      to spot that kinda thing. looking back I wonder if perhaps we should
      just stop using sa_len, would certainly avoid this ifdef.
      
      should fix building on freebsd
      d7ba51de
    • Harald Sitter's avatar
      do not disable getifaddrs use because of sa_len availablility · 4582053b
      Harald Sitter authored
      Summary:
      this fixes ipv6 support in nic listing. well, kinda.
      
      the getifaddrs branch is the only one supporting ipv6 properly
      so it seems silly to use the ioctl branch just because sa_len
      is not available.
      simply wrap around getnameinfo and either use sa_len or
      determine the length based on the socket family (as seen on the
      getifaddrs manpage). this way the branch compiles fine without
      sa_len and is fine to use by just about all glibc systems thus
      giving us ipv6 support on just about every glib system.
      
      BUG: 313892
      BUG: 277208
      FIXED-IN: 5.18.0
      
      Test Plan: built on neon. shows ipv6 addresses
      
      Reviewers: ngraham
      
      Reviewed By: ngraham
      
      Subscribers: plasma-devel
      
      Tags: #plasma
      
      Differential Revision: https://phabricator.kde.org/D26729
      4582053b
  13. 30 Jan, 2020 2 commits
  14. 22 Jan, 2020 3 commits
    • Harald Sitter's avatar
      tooltipmanager: use qstyle wakeup delay instead of hardcoding · 903df09d
      Harald Sitter authored
      Summary:
      aids in making kinfocenter and systemsettings behave with the
      same delay. plus not hardcoding things is always cool
      
      BUG: 393722
      FIXED-IN: 5.18.0
      
      Test Plan: builds, tooltip shows up, qdebugging says timeout is now 700
      
      Reviewers: mart
      
      Reviewed By: mart
      
      Subscribers: mart, ngraham, bcooksley, plasma-devel
      
      Tags: #plasma
      
      Differential Revision: https://phabricator.kde.org/D26733
      903df09d
    • Harald Sitter's avatar
      override picutils' error logger; it has an exit() call · e677d2cc
      Harald Sitter authored
      Summary:
      instead simply run the output through qwarning as a cheap
      solution.
      
      FIXED-IN: 5.18.0
      BUG: 382979
      
      Test Plan: builds, dont have a device without pci
      
      Reviewers: ngraham
      
      Reviewed By: ngraham
      
      Subscribers: plasma-devel
      
      Tags: #plasma
      
      Differential Revision: https://phabricator.kde.org/D26807
      e677d2cc
    • Harald Sitter's avatar
      parse xorg version for nicer display · 47ec0bda
      Harald Sitter authored
      Summary:
      instead of just printing a random integer lets try to print a
      nice looking version string. since I couldn't find
      cross-implementation documentation on what the format of the
      int is, I am going to assume it is vendor specific and guard
      the xorg specific parsing behind a guard based on the vendor
      string containig x.org
      
      BUG: 344038
      FIXED-IN: 5.18.0
      
      Test Plan: version looks nice in xorg module
      
      Reviewers: mart
      
      Reviewed By: mart
      
      Subscribers: plasma-devel
      
      Tags: #plasma
      
      Differential Revision: https://phabricator.kde.org/D26730
      47ec0bda
  15. 20 Jan, 2020 4 commits
    • Harald Sitter's avatar
      do not show "copy in english" when using C as locale · 87d793d1
      Harald Sitter authored
      Summary:
      qlocale differentiates LANG=C from LANG=en_US for us it is
      though as the native untranslated strings are english, so when
      kinfocenter is invoked with C it will effectively be == en_US.
      with that in mind we don't need the copy in english button when
      using C.
      
      BUG: 403368
      FIXED-IN: 5.18.0
      
      Test Plan: LANG=C.UTF-8 no longer causes the button to be visible
      
      Reviewers: mart, ngraham
      
      Reviewed By: mart, ngraham
      
      Subscribers: plasma-devel
      
      Tags: #plasma
      
      Differential Revision: https://phabricator.kde.org/D26792
      87d793d1
    • Harald Sitter's avatar
      find org.kde.kirigami as runtime dependency · ced4536e
      Harald Sitter authored
      Summary:
      without it the energy and indexer modules won't work.
      
      BUG: 403011
      FIXED-IN: 5.18.0
      
      Test Plan: org.kde.kirigami listed as runtime dep and reported missing when not installed
      
      Reviewers: broulik, ngraham
      
      Reviewed By: ngraham
      
      Subscribers: plasma-devel
      
      Tags: #plasma
      
      Differential Revision: https://phabricator.kde.org/D26789
      ced4536e
    • Harald Sitter's avatar
      do not needlessly run through Solid::Device::allDevices · d2200025
      Harald Sitter authored
      Summary:
      allDevice is bugging out and having duplicated entires
      https://bugs.kde.org/show_bug.cgi?id=416495
      this results in garbage in our tree view.
      on top of that running through allDevices is entirely pointless
      as we can directly construct a Device from the udi we got which
      is not only faster, it also lets us bypass the aformentioned
      bug since the udi is still unique it's only the objects that
      get duped
      
      BUG: 399746
      FIXED-IN: 5.18.0
      
      Test Plan: repated plug/unplug of usb drive doesn't cause bogus device entries
      
      Reviewers: apol, ngraham
      
      Reviewed By: apol, ngraham
      
      Subscribers: plasma-devel
      
      Tags: #plasma
      
      Differential Revision: https://phabricator.kde.org/D26788
      d2200025
    • Harald Sitter's avatar
      remove uselss boilerplate line · 06d71637
      Harald Sitter authored
      GIT_SILENT
      06d71637
  16. 19 Jan, 2020 1 commit
  17. 17 Jan, 2020 1 commit
    • Harald Sitter's avatar
      make sure Solid::Battery is not deleted from QML · 764fce4a
      Harald Sitter authored
      Summary:
      Battery objects are casted DeviceInterface objects and those are
      owned by Solid. deleting them outside solid means they will end
      up as dangling pointers inside Solid's global static objects.
      
      when switching away from the energy KCM the qml engine would
      get cleaned up as part of the KCM destruction, qml would then
      sweep up the Battery object and corrupt the solid internal
      pointers.
      
      to prevent this, explicitly mark Battery objects we give to QML
      as owned on the c++ side.
      
      BUG: 413003
      FIXED-IN: 5.18.0
      
      Test Plan:
      open kinfocenter
      switch to energy
      switch away
      switch to energy
      no crash
      
      Reviewers: broulik, davidedmundson
      
      Reviewed By: davidedmundson
      
      Subscribers: plasma-devel
      
      Tags: #plasma
      
      Differential Revision: https://phabricator.kde.org/D26725
      764fce4a
  18. 16 Jan, 2020 2 commits
  19. 07 Jan, 2020 2 commits
  20. 06 Jan, 2020 2 commits
  21. 04 Jan, 2020 1 commit
  22. 30 Dec, 2019 1 commit
  23. 26 Dec, 2019 1 commit
  24. 02 Dec, 2019 1 commit
  25. 26 Nov, 2019 1 commit