1. 13 Aug, 2008 1 commit
  2. 11 Aug, 2008 1 commit
  3. 06 Jul, 2008 4 commits
  4. 06 Jun, 2008 1 commit
  5. 26 May, 2008 2 commits
  6. 23 May, 2008 1 commit
  7. 10 May, 2008 2 commits
  8. 04 May, 2008 1 commit
  9. 01 May, 2008 1 commit
  10. 17 Apr, 2008 1 commit
    • Michael Ploujnikov's avatar
      fixed misc/stats.c:399: warning: comparison between signed and unsigned in... · 47240a15
      Michael Ploujnikov authored
      fixed misc/stats.c:399: warning: comparison between signed and unsigned in function '__stats_TimerClean'
      
      Based on usage of i_id, other stats_* functions and my opinion i_id should be unsigned int instead of int, but I might be missing the bigger picture related to changing public APIs. Please, let me know how to correctly deal with such warnings.
      47240a15
  11. 15 Apr, 2008 3 commits
  12. 14 Apr, 2008 1 commit
  13. 25 Mar, 2008 1 commit
  14. 09 Feb, 2008 1 commit
  15. 24 Jan, 2008 2 commits
  16. 12 Dec, 2007 1 commit
  17. 16 Feb, 2007 1 commit
  18. 22 Dec, 2006 1 commit
    • Clément Stenac's avatar
      Add facilities to report and enrich error messages. · 65d21833
      Clément Stenac authored
      The idea is that a given error should be reported only once to the user.
      At the moment, for example, we can get:
      - "no suitable access module" (printed by module_Need)
       - "unable to create access" (printed by input/access)
       - "unable to open stream" (printed by input/input)
       - ...
      
      The facilities provided here are:
      * msg_StackSet( code, message ) : "throw" a new error
      * msg_StackAdd( message ) : Append a message allowing to trace the message
      * msg_StackMsg() : Retrieve the message
      
      So, in the previous example, module_Need would msg_StackSet( code, "no
      suitable access module"), then input/input would add "unable to create
      access" and finally, input/input can print the "unable to open stream:
      unable to create access: no suitable module" message.
      
      The three functions are context-free, you don't need a vlc_object, so
      that we can add verbose error reporting everywhere.
      
      65d21833
  19. 03 Dec, 2006 1 commit
    • Clément Stenac's avatar
      A bit of vlc/libvlc cleanup: · fbf4c806
      Clément Stenac authored
      * libvlc sources and plugins now define __LIBVLC__ and headers exported to
        plugins (ie, include/vlc_*.h) check for this
      * libvlc sources use the value of the "libvlc" component in vlc-config instead
        of vlc
      * Apply flags directly to "libvlc" in configure.ac - might temporarily break win32 :)
      fbf4c806
  20. 27 Oct, 2006 1 commit
  21. 21 Oct, 2006 1 commit
  22. 18 Sep, 2006 1 commit
  23. 03 Sep, 2006 1 commit
  24. 29 Jul, 2006 1 commit
  25. 15 Jun, 2006 1 commit
  26. 25 May, 2006 1 commit
  27. 14 May, 2006 1 commit
    • Clément Stenac's avatar
      Merge back branch 0.8.6-playlist-vlm to trunk. · ed0b72e3
      Clément Stenac authored
      What is currently broken:
       * Some playlist demuxers (shout, dvb and pls)
       * DAAP
       * BeOS playlist
       * GPE playlist, I suppose
      
      What has some trouble:
       * Meta handling in several demuxers (most notably TS)
       * Skins2 playlist (doesn't refresh correctly)
       * OS X playlist (see bigben's latest commits)
      ed0b72e3
  28. 30 Apr, 2006 1 commit
  29. 29 Apr, 2006 1 commit
  30. 04 Mar, 2006 2 commits
  31. 19 Feb, 2006 1 commit