1. 02 Dec, 2020 1 commit
  2. 25 Nov, 2020 1 commit
  3. 21 Nov, 2020 1 commit
  4. 08 Nov, 2020 1 commit
  5. 07 Nov, 2020 2 commits
  6. 06 Nov, 2020 1 commit
    • Jurica Vukadin's avatar
      sftp: Fix build failure with old libssh versions · c973cfaa
      Jurica Vukadin authored
      Building kio-sftp fails on Ubuntu 18.04:
      kio-extras/sftp/kio_sftp.cpp:1280:1: error: non-void function does not return a value in all control paths [-Werror,-Wreturn-type]
      }
      ^
      1 error generated.
      make[2]: *** [sftp/CMakeFiles/kio_sftp.dir/kio_sftp.cpp.o] Error 1
      make[1]: *** [sftp/CMakeFiles/kio_sftp.dir/all] Error 2
      make[1]: *** Waiting for unfinished jobs....
      make: *** [all] Error 2
      
      SFTPInternal::openConnection() version for libssh < 0.8.3 is missing a
      return at the end of the function. Add it.
      c973cfaa
  7. 02 Nov, 2020 1 commit
    • Harald Sitter's avatar
      smb: revise auth behavior as per upstream suggestions · be979be8
      Harald Sitter authored
      - We no longer attempt to auto-anonymous as that can increase the failed
      logon count for no good reason. As a silly side effect it'd also mean
      that servers that actually support anonymous login would always be used
      anonymously and the user then has no visible way to actually login short
      of editing the URI to smb://username@host
      - Domain field is gone now. Folded into username as per UPN foramt
      - Anonymous field is also gone now. Also folded into username
      
      This largely roots in the fact that windows logons are a range of
      formats, so upstream rightly suggested that they all should go through
      the same username field (which is also how this is done on windows).
      This also streamlines the auth dialog a lot and removes a bit of excess
      code.
      https://docs.microsoft.com/en-us/windows/win32/secauthn/user-name-formats
      
      Anonymous/guest login actually is also implemented thusly by either
      using empty or invalid usernames (kinda depends on how the server,
      specifically samba, is configured anyway).
      
      To communicate these options to the user we use a new feature in KF5
      that allows us to set a fairly visible hint on the username field to
      describe in greater detail what the username may be.
      
      I've tested this against my windows server 2019 setup and all the login
      options described work as expected.
      be979be8
  8. 31 Oct, 2020 1 commit
  9. 30 Oct, 2020 2 commits
  10. 29 Oct, 2020 2 commits
  11. 28 Oct, 2020 1 commit
  12. 25 Oct, 2020 2 commits
  13. 24 Oct, 2020 1 commit
  14. 20 Oct, 2020 1 commit
  15. 19 Oct, 2020 4 commits
  16. 18 Oct, 2020 2 commits
    • Stefan Brüns's avatar
      Filter out resource forks when filtering for images · efd12724
      Stefan Brüns authored
      The pages in a comic book are just a set of images, ordered by filename.
      When there is a resource fork (as created by MacOS when using the
      built-in ZIP support), the metadata files (which have the same suffix as
      the corresponding image files) are sorted first.
      
      For e.g a two page book the files `book_01.png`, `book_02.png` are sorted
      after and hidden by `__MACOSX/._book_01.png`, `__MACOSX._book_02.png`.
      efd12724
    • Stefan Brüns's avatar
      Avoid pointless probing for nonexisting thumbnail images · ef76b22b
      Stefan Brüns authored
      QImage::load(<path>) tries to find a fallback image if the file does
      not exist, by appending extensions of all supported file formats, as
      QImageReader defaults to format autodetection.
      
      Check if the file exists, and only then try to load it.
      ef76b22b
  17. 14 Oct, 2020 3 commits
    • Harald Sitter's avatar
      smb: throw out discovery timeout · c2e576ab
      Harald Sitter authored
      it was only ever a safety net and shouldn't practically be necessary.
      with the dust around the new discovery system having settled we should
      probably deal with potentially getting stuck on discovery instead of
      aborting the discovery through the safety net and hiding the defect.
      
      also, technically speaking discovery could take longer than N seconds
      since both WSD and NMB are broadcast systems without local caching
      daemon like avahi, so they can be affected by network latency much more
      than DNSSD
      c2e576ab
    • Harald Sitter's avatar
      Merge branch 'release/20.08' · 77517010
      Harald Sitter authored
      77517010
    • Harald Sitter's avatar
      smb: no discovery cap on file listing · c58fdc63
      Harald Sitter authored
      File and share listing can take a very long time as they are IO bound on
      the remote, so if one is listing a large directory on a HDD it may well
      take minutes. Equally sending huge listings over the wire may take a
      while if not enough throughput is available. So don't cap the execution
      of these discoveries at all.
      
      Server discovery retains the timeout, mostly because I don't want to
      remove it in 20.08 though an argument could be made that we should trust
      in the reliability of the discovery stacks to not get stuck on random
      unforseen misbehavior of servers.
      
      BUG: 427644
      FIXED-IN: 20.08.3
      c58fdc63
  18. 12 Oct, 2020 2 commits
  19. 11 Oct, 2020 2 commits
  20. 10 Oct, 2020 4 commits
  21. 09 Oct, 2020 1 commit
  22. 05 Oct, 2020 2 commits
  23. 03 Oct, 2020 1 commit
  24. 25 Sep, 2020 1 commit