1. 26 Jan, 2021 1 commit
    • Andre Heinecke's avatar
      Fix Windows UNC paths for file encryption · 86cf610a
      Andre Heinecke authored
      With Qt 5.13.2 (and still there in 5.15.0)
      there was a regression that QTemporaryFile on Windows
      UNC drives (unmounted Samba shares) have a bad filename even
      though the file is correctly created. As the fix in Qt for
      this is not trivial this adds a workaround to fix up the path name
      and can probably be removed in a future version.
      
      GnuPG-Bug-Id: T5216
      86cf610a
  2. 05 Nov, 2020 1 commit
  3. 13 Aug, 2020 1 commit
  4. 16 Jan, 2020 1 commit
  5. 06 Mar, 2019 1 commit
    • Andre Heinecke's avatar
      Add output process error handling · 20d29e40
      Andre Heinecke authored
      This adds error handling for the output process similar to:
      0ebc73bb
      
      Previously output process errors were ignored and the decryption
      was shown as successful even if a corrupted tar archive was
      extracted. This lead to mysterious missing files after
      the decryption of corrupted archives.
      
      GnuPG-Bug-Id: T4332
      20d29e40
  6. 26 Oct, 2018 1 commit
  7. 04 Jul, 2018 1 commit
    • Andre Heinecke's avatar
      Handle no MDC for legacy algos gracefully · 9e5fa865
      Andre Heinecke authored
      With GPGME 1.11.2 we can detect if an MDC error occured
      because of an old algorithm. This is common enough that
      we need a proper handling.
      
      It is now detected, the user is shown a notification and
      gets the option to "Force decryption" so that the
      user could then ideally re-encrypt the file with a modern
      algo.
      
      GnuPG-Bug-Id: T4038
      9e5fa865
  8. 17 Jan, 2018 1 commit
  9. 08 Nov, 2017 1 commit
    • Daniel Vrátil's avatar
      Fix detached S/MIME signature verification · 2edf2b81
      Daniel Vrátil authored
      GPG cannot properly distinguish whether a file contains a detached
      or an opaque S/MIME signature. So we now first try to check if a file
      with the same base name as the S/MIME signature file exists and treat
      the signature as detached. If no such file exists, we assume opaque
      signature.
      
      This change also re-orders the way the tasks are scheduled to make sure
      that decryption happens before signature verification. For that we need
      to "pipe" output from the decryption task into the verification task.
      Since holding the decrypted content in-memory could be very costly in
      case of a large file, this change introduces Crypto::Input that can
      work on Crypto::FileOutput in a way that the FileOutput signals to the
      Input that the file now exists.
      
      Differential Revision: https://phabricator.kde.org/D8376
      2edf2b81
  10. 22 Aug, 2017 1 commit
    • Daniel Vrátil's avatar
      Replace assert() with Q_ASSERT() · f4ce2ec7
      Daniel Vrátil authored
      Fixes 'lambda-expression in unevaluated context' error in some places and generally
      switches to Q_ASSERT, which is what we should be using in a Qt app anyway.
      f4ce2ec7
  11. 04 Jun, 2017 1 commit
  12. 25 Feb, 2017 1 commit
  13. 10 Jan, 2017 1 commit
  14. 04 Jan, 2017 1 commit
  15. 12 Oct, 2016 1 commit
    • Daniel Vrátil's avatar
      Port Kleopatra to new gpgme 1.7 · 920027cc
      Daniel Vrátil authored
      This also includes lots of boost::shared_ptr -> std::shared_ptr
      changes, but there are still places where boost is used. We should
      definitely do a second round of clean up later on.
      920027cc
  16. 07 Mar, 2016 1 commit
  17. 13 Oct, 2015 1 commit
  18. 19 Sep, 2015 1 commit
  19. 16 Sep, 2015 1 commit
  20. 14 Sep, 2015 1 commit
  21. 08 Aug, 2015 1 commit
  22. 29 Jul, 2015 1 commit
  23. 03 May, 2015 1 commit
  24. 29 Apr, 2015 1 commit
  25. 03 Nov, 2014 1 commit
  26. 04 Jun, 2014 1 commit
  27. 18 Dec, 2013 2 commits
  28. 20 Aug, 2013 2 commits
    • Andre Heinecke's avatar
      Fix ProcessStdInOutput finalize · 529d3b7e
      Andre Heinecke authored
          This makes sure that the data is completly written
          to the Process before waiting for the process
          to finish. If this is not done the output will be written
          in small chunks trough the eventloop causing an uncessary
          delay before the process on which we wait has even a chance
          to work.
          This delay is mainly noticabe on Windows where it can
          take ~30 seconds to write out a 10MB file in the 512 byte chunks
          that are provided by gpgme in QProcess::write calls.
          Also fixes a problem where sometimes the last chunk was not
          sent before the Output process closed.
      529d3b7e
    • Andre Heinecke's avatar
      Add uiserver support for OUTPUT --binary option · 7a14561d
      Andre Heinecke authored
          Since r393a947 gpgme specifies the --binary option to be
          accepted in an uiserver OUTPUT command. This option should
          overwrite the specified default behavior for uiservers to
          create ASCII armored output.
          The uiserver uses the encrypt and sign email tasks of kleopatra
          so the option needs to be handled there.
      7a14561d
  29. 09 Aug, 2013 1 commit
    • Andre Heinecke's avatar
      Fix ProcessStdInOutput finalize · 6d3f5bc9
      Andre Heinecke authored
          This makes sure that the data is completly written
          to the Process before waiting for the process
          to finish. If this is not done the output will be written
          in small chunks trough the eventloop causing an uncessary
          delay before the process on which we wait has even a chance
          to work.
          This delay is mainly noticabe on Windows where it can
          take ~30 seconds to write out a 10MB file in the 512 byte chunks
          that are provided by gpgme in QProcess::write calls.
          Also fixes a problem where sometimes the last chunk was not
          sent before the Output process closed.
      6d3f5bc9
  30. 01 Aug, 2013 1 commit
    • Andre Heinecke's avatar
      Add uiserver support for OUTPUT --binary option · 71aad808
      Andre Heinecke authored
          Since r393a947 gpgme specifies the --binary option to be
          accepted in an uiserver OUTPUT command. This option should
          overwrite the specified default behavior for uiservers to
          create ASCII armored output.
          The uiserver uses the encrypt and sign email tasks of kleopatra
          so the option needs to be handled there.
      71aad808
  31. 27 Jul, 2013 1 commit
  32. 23 Jul, 2013 1 commit
  33. 26 May, 2012 1 commit
  34. 18 Nov, 2010 1 commit
  35. 26 Oct, 2010 2 commits
  36. 23 Sep, 2010 1 commit
  37. 07 Sep, 2010 1 commit
    • Thomas McGuire's avatar
      SVN_MERGE: · b09634ef
      Thomas McGuire authored
      Merged revisions 1171031 via svnmerge from 
      svn+ssh://tmcguire@svn.kde.org/home/kde/branches/kdepim/enterprise4/kdepim
      
      ........
        r1171031 | mutz | 2010-09-02 16:03:25 +0200 (Thu, 02 Sep 2010) | 6 lines
        
        Add Output::errorString().
        
        It defaults to QIODevice::errorString(), but is overridden for QProcess to
        return the process' stderr.
        
        Probably not working yet.
      ........
      
      svn path=/trunk/KDE/kdepim/kleopatra/; revision=1172596
      b09634ef