Menu

crash when closing windows in 0.4.0

Help
2006-10-20
2013-01-15
  • Joshua J. Wingstrom

    Version 0.4.0 is great.  I really like it, but when I have multiple windows open, and I close one, it crashes.  I am running Debian testing and installed RKWard from experimental.  I tried to post a bug report at bugs.kde.org per the instructions in the about dialog box, however, I could not find rkward listed there.  Is anyone else having this problem?

    Thank-you.

     
    • Thomas Friedrichsmeier

      By "when I have multiple windows open", you mean "detached" windows (i.e. not in the tabbed area)? In that case, I've found, and fixed the bug in the development version. If you get crashes while closing windows under any under circumstances, please write back.

      Oh, indeed, those instructions are entirely misleading. They are part of the standard KDE-application help menu, but RKWard is not an official part of KDE. I'll have to figure out, how to remove/replace those parts. This location for reporting bugs is fine. Another option would have been the mailing list, or of course the bug tracker:
      https://sourceforge.net/tracker/?group_id=50231&atid=459007

      Thanks for reporting these issues!

       
    • Joshua J. Wingstrom

      Thomas,

      You are very welcome.  The least I can do is report this stuff.  You've made my life a lot easier by writing this.

      By "when I have multiple windows open" I mean when I have multiple tabbed windows open.  For example, see:

      http://navet.ics.hawaii.edu/~wingstro/rkward-ss.png

      If I press ctrl+w right now, rkward will crash.

      Thanks!

       
      • Thomas Friedrichsmeier

        Hi Joshua,

        I see. Unfortunately, that scenario works just fine, here, so I don't know where this is coming from. Some more questions to try to track this down:

        1) Does this happen every time? Or does it happen just sometimes? If the latter, can you figure out a pattern, when it happens?

        2) Does it only happen, when you have several windows open, and try to close one, or also if there is just a single window, and you close that?

        3) Could you try starting rkward from the console, and provoke the crash. Are there any messages on the console?

        4) Esp., if 3) does not show anything, could you start rkward with
        rkward --debug-level 5 2> logfilename
        Startup will be noticeably slower, and a lot of debug output should go into logfilename. Could you zip that up, and - if the .gz ends up smaller than say 100Kb, post to rkward-devel@lists.sourceforge.net, if it's larger than that, send it to me via private mail, or put it on the webserver.

        5) Are you using a 64bit system by any chance?

        Thanks!

         
        • Thomas Friedrichsmeier

          Just to make sure: As part of 4), also try to provoke the crash. After all, that is the point of interest.

           
        • Joshua J. Wingstrom

          1) Does this happen every time? Or does it happen just sometimes? If the latter, can you figure out a pattern, when it happens?

          It happens every time.

          2) Does it only happen, when you have several windows open, and try to close one, or also if there is just a single window, and you close that?

          It only happens if there are two or more windows open.

          3) Could you try starting rkward from the console, and provoke the crash. Are there any messages on the console?

          Two interesting ones:
          Debug-flags as decimal: 8191
          kdecore (KAction): WARNING: KAction::insertKAccel( kaccel = 0x10197278 ): KAccel object already contains an action name "window_close"

          and:
          Segmentation fault

          4) Esp., if 3) does not show anything, could you start rkward with
          rkward --debug-level 5 2> logfilename

          http://navet.ics.hawaii.edu/~wingstro/rkward-debug.log.gz

          5) Are you using a 64bit system by any chance?

          Nope, Mac Laptop:
          $ cat /proc/cpuinfo
          processor       : 0
          cpu             : 7447A, altivec supported
          clock           : 1666.666000MHz
          revision        : 0.5 (pvr 8003 0105)
          bogomips        : 33.15
          timebase        : 8320000
          platform        : PowerMac
          machine         : PowerBook5,8
          motherboard     : PowerBook5,8 MacRISC3 Power Macintosh
          detected as     : 287 (PowerBook G4 15")
          pmac flags      : 00000019
          L2 cache        : 512K unified
          pmac-generation : NewWorld

          Thank-you!

           
    • Thomas Friedrichsmeier

      Hi Joshua,

      thanks for all the info. Unfortunately, I still don't quite understand, why this is happening. The crash is definitely inside RKWorkplaceView::setPage () (and more precisely, in the second half of that, after the debug statement), but I fail to see, what exactly is going wrong there. There is a small inaccuracy in the debug statement itself, which prevents compilation on 64 bit systems, but a) it compiles for you, and b) Other calls to setPage () (e.g. when you're manually switching between tabs) seem to work fine.

      How to track this down? Maybe, as a small attempt, you could try to follow the advice given here: https://sourceforge.net/forum/message.php?msg_id=3974958 to clean up the inaccuracy. Does this affect the bug in any way?

      If this does not change anything (I suspect it won't), could you try commenting out (adding "//" at the beginning of those lines) lines 137, 139, 145, and 146 one by one, recompile, and test each time? One of those must be the one causing trouble on your machine, so the crash should go away if you comment out the right one (you will get some other minor bugs instead). Can you identify the line causing trouble this way?

      Sorry, I'm just asking more questions instead of providing a fix, but unfortunately, I simply can not test this bug myself, as it does not occur, here. Probably this is specific to your platform / CPU in some way or another.

      Thanks!

       
    • Thomas Friedrichsmeier

      Hi!

      RKWard 0.4.1 should now be in debian unstable (or become available real soon, now). While I did not do anything to specifically address this problem, maybe the 64bit fix affects this issue.

      Could you try installing 0.4.1, and test whether the bug is still there, or not?

      Thanks!