Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Supported by

Bug report OS 0.27~pre15

edited October 2012 in OpenSesame

Bug: When running an experiment with the xpyriment backend, the experiment frequently aborts with the following error message and stack trace:

Traceback (most recent call last):
  File "/home/daniel/Ubuntu One/opensesame-0.27~pre15/libqtopensesame/qtopensesame.py", line 1343, in run_experiment
    exp.run()
  File "/home/daniel/Ubuntu One/opensesame-0.27~pre15/libopensesame/experiment.py", line 303, in run
    self.items[self.start].run()
  File "/home/daniel/Ubuntu One/opensesame-0.27~pre15/libopensesame/sequence.py", line 60, in run
    self.experiment.items[item].run()
  File "/home/daniel/Ubuntu One/opensesame-0.27~pre15/libopensesame/loop.py", line 160, in run
    _item.run()
  File "/home/daniel/Ubuntu One/opensesame-0.27~pre15/libopensesame/sequence.py", line 60, in run
    self.experiment.items[item].run()
  File "/home/daniel/Ubuntu One/opensesame-0.27~pre15/libopensesame/loop.py", line 159, in run
    if _item.prepare():
  File "/home/daniel/Ubuntu One/opensesame-0.27~pre15/libopensesame/sequence.py", line 120, in prepare
    if not self.experiment.items[_item].prepare():
  File "/home/daniel/Ubuntu One/opensesame-0.27~pre15/plTraceback (most recent call last):
  File "/home/daniel/Ubuntu One/opensesame-0.27~pre15/libqtopensesame/qtopensesame.py", line 1343, in run_experiment
    exp.run()
  File "/home/daniel/Ubuntu One/opensesame-0.27~pre15/libopensesame/experiment.py", line 303, in run
    self.items[self.start].run()
  File "/home/daniel/Ubuntu One/opensesame-0.27~pre15/libopensesame/sequence.py", line 60, in run
    self.experiment.items[item].run()
  File "/home/daniel/Ubuntu One/opensesame-0.27~pre15/libopensesame/loop.py", line 160, in run
    _item.run()
  File "/home/daniel/Ubuntu One/opensesame-0.27~pre15/libopensesame/sequence.py", line 60, in run
    self.experiment.items[item].run()
  File "/home/daniel/Ubuntu One/opensesame-0.27~pre15/libopensesame/loop.py", line 159, in run
    if _item.prepare():
  File "/home/daniel/Ubuntu One/opensesame-0.27~pre15/libopensesame/sequence.py", line 120, in prepare
    if not self.experiment.items[_item].prepare():
  File "/home/daniel/Ubuntu One/opensesame-0.27~pre15/plugins/fixation_dot/fixation_dot.py", line 77, in prepare
    self.c.fixdot(self._x, self._y)
  File "/home/daniel/Ubuntu One/opensesame-0.27~pre15/openexp/_canvas/xpyriment.py", line 155, in fixdot
    stim = stimuli.Circle(16, colour=color, position=c2p((x,y)))
AttributeError: 'module' object has no attribute 'Circle'ugins/fixation_dot/fixation_dot.py", line 77, in prepare
    self.c.fixdot(self._x, self._y)
  File "/home/daniel/Ubuntu One/opensesame-0.27~pre15/openexp/_canvas/xpyriment.py", line 155, in fixdot
    stim = stimuli.Circle(16, colour=color, position=c2p((x,y)))
AttributeError: 'module' object has no attribute 'Circle'

(Platform: Ubuntu 12.04; but it also occurs under Windows 7)

Comments

  • edited 9:01AM

    Thanks for your report! The crash is likely due to the fact that you use an older version of Expyriment, in which there was no Circle stimulus. You're probably running from source on Windows 7, right?

    On Ubuntu, you can get the latest version of Expyriment and OpenSesame from the smathot/cogscinl ppa:

    sudo add-apt-repository ppa:smathot/cogscinl
    sudo apt-get update
    sudo apt-get install expyriment opensesame

    On Windows it's probably easiest to use the pre-release package or get the latest Expyriment source from the Ubuntu PPA (trough View package details).

    Cheers!

    There's much bigger issues in the world, I know. But I first have to take care of the world I know.
    cogsci.nl/smathot

  • edited October 2012

    I'm running on Ubuntu at the moment and your suggestion fixed the problem. Thanks!
    I'll see how this fares on Windows a bit later.

    Is the code of expyriment on code.google.com not the latest?

  • edited October 2012

    Platform: Ubuntu 12.04

    Bug: Sometimes when an experiment is started in windowed mode and you switch to another window (outside OS) after which you switch back to the experiment window again, it does not register any key presses anymore (including ESC). You have to force quit OpenSesame or kill the experiment window with xkill

    EDIT: It even happens at random at the start-up of the experiment. Without switching windows (or releasing the focus from the experiment window) the experiment simply does not register responses and gets stuck at the instructions screen. You can only exit by killing the window the hard way in this situation.


    Bug: OpenSesame occasionally hangs when opening an experiment. At the moment you select a file and click on the Load button, OpenSesame freezes for a minute and then exits, after which the "send error feedback" window of Ubuntu appears

  • edited 9:01AM
    Is the code of expyriment on code.google.com not the latest?

    Not the 0.5.2 released version, no. OpenSesame relies on some new functionality. You can checkout the latest source, though, (if you have mercurial installed) with:

    hg clone https://code.google.com/p/expyriment/

    Regarding the bugs: That doesn't sound too good. Thanks I'll check it out!

    There's much bigger issues in the world, I know. But I first have to take care of the world I know.
    cogsci.nl/smathot

  • edited 9:01AM
    Bug: Sometimes when an experiment is started in windowed mode and you switch to another window (outside OS) after which you switch back to the experiment window again, it does not register any key presses anymore (including ESC). You have to force quit OpenSesame or kill the experiment window with xkill

    I haven't experienced this myself. Can you maybe trace it down further? I.e. does it happen with all backends? Is there a specific experiment that triggers the problem. Did this also happen with 0.26? I'm thinking it may have something to do with compiz, the effects layer used by Ubuntu. If you replace compiz by a non-compositing window manager, by running

    metacity --replace

    does it still happen?

    Bug: OpenSesame occasionally hangs when opening an experiment. At the moment you select a file and click on the Load button, OpenSesame freezes for a minute and then exits, after which the "send error feedback" window of Ubuntu appears

    Hmm, I've never seen this either. If you start OpenSesame in a terminal, do you get any useful output? And, again. does it happen with a specific experiment, or does it appear to be completely random?

    There's much bigger issues in the world, I know. But I first have to take care of the world I know.
    cogsci.nl/smathot

  • edited October 2012

    Thanks for you reply

    I haven't experienced this myself. Can you maybe trace it down further? I.e. does it happen with all >backends? Is there a specific experiment that triggers the problem. Did this also happen with 0.26? >I'm thinking it may have something to do with compiz, the effects layer used by Ubuntu. If you >replace compiz by a non-compositing window manager, by running

    >

    metacity --replace

    >

    does it still happen?

    I will try that. Will that however interfere with other Ubuntu windows too? As in, will it disable the compiz effects for desktop switching and such? As far as I noticed, this didn't happen with 0.26 (and the hang-ups on load did neither). Last grading session I loaded many experiments and that all went fine. In OS 0.27, it happens with legacy and expyriment backends (haven't tried psychopy)

    Hmm, I've never seen this either. If you start OpenSesame in a terminal, do you get any useful >output? And, again. does it happen with a specific experiment, or does it appear to be completely >random?

    I have been grading assignments all day, so I think about 90 files (21 students x 4 different experiments) "passed the revue". In 1 out of 10-15 file loads I got this hang-up and I haven't noticed a pattern yet. It hence appears to be completely random. I'll try to see what output I can get in a terminal; until now this has only been "Segmentation fault"

  • edited October 2012
    Will that however interfere with other Ubuntu windows too? As in, will it disable the compiz effects for desktop switching and such?

    Yes, it will. But temporarily, just to find out if compiz causes the trouble. You can always switch back to compiz by logging out/in or by running

    compiz --replace

    I'm not sure what the cause of these issues could be, and it's difficult to pin down if they don't occur on my system... Please let me know if you can get your hands on some potentially useful debugging output!

    There's much bigger issues in the world, I know. But I first have to take care of the world I know.
    cogsci.nl/smathot

  • edited October 2012

    Platform: Win XP

    Bug (sort of): The form_consent item works now, but at a resolution of 1024x768, the "I have read and understood the information above" checkbox label overlaps with the main text (the word uncomfortable). Additionally, when you run the experiment in full screen, you don't get to see the consent form at all; you only get a black screen with a mouse cursor. Once you click anywhere in that black screen with your mouse, you get to see the consent form for a couple of frames, after which the screen goes black again.


    Bug: The not listening to responses from the very start of the experiment (earlier described for Ubuntu) also occurs on Windows XP. I am stuck in the first instructions screen and it doesn't continue when pressing space. I have to kill OpenSesame in the task manager to stop the experiment (or close the command line box from which I am running). I therefore don't think this problem is caused by compiz in Ubuntu. I also do not get to see any useful debugging output in the console (running with --debug) when this lock out happens. The next couple of times the experiment is rerun, it works fine from beginning to end, so it is unlikely an infinite loop in my code.


    Bug: This is kind of a major one: when an experiment successfully finishes or you press ESC to exit during the experiment, the experiment window closes, after which your mouse cursor gets 'stuck' in the middle of the screen. You are unable to move your mouse cursor and have to press CTRL-ALT-DEL to get control back (and just press 'cancel' in the windows security dialog). Clicking still seems to be registered while the mouse can't be moved. So far, this has happened after every run of various experiments (so no random occurences here)

    Edit: Does not seem to occur on Windows 7; only Windows XP

    Edit 2: Seems only to occur after experiments that were run in windowed mode. After a fullscreen experiment is finished the problem does not occur.

  • edited 9:01AM

    Thanks again! I'm going to look into it. Just one quick question: Are you running from source (and, if so, which versions of the relevant modules are you using), or do you use the pre-release binary?

    There's much bigger issues in the world, I know. But I first have to take care of the world I know.
    cogsci.nl/smathot

  • edited October 2012

    I'm running from source. modules() gives the following output:

    >>> modules()
    OpenSesame 0.27~pre15
    Python 2.7.2
    Expyriment  (Revision ; Python 2.7.2)
    PyAudio not available
    PyGame 1.9.2pre
    PyOpenGL 3.0.2a6
    PyQt 4.8.6
    PySerial 2.6
    PsychoPy 1.74.02
    Pyglet 1.1.4
    

    I see that my version of PyQt4 is hopelessly outdated. I updated it to 4.9.5, but the problem still occurs... The strange thing is that it only occurs after an experiment exits that was run with the testrun button. Normal windowed or fullscreen experiments show no problems of this kind.

    Edit: It does also occur after an experiment that was run in normal windowed mode exits. So with all windowed modes the problem occurs, but with the fullscreen mode it doesn't

  • edited October 2012

    The problem of the frozen mouse cursor also occurs when I run OpenSesame 0.26 from source (the distributed version runs fine). I therefore think the problem is specific to the configuration of my machine and not a bug in OpenSesame 0.27.

  • edited 9:01AM

    Platform: Win XP

    Bug: Trying to run (any) experiment with the psychopy backend results in the following error as soon as the experiment starts (you do not get to see the window):

    openexp.sampler._legacy.init_sound(): sampling freq = 48000, buffer size = 512
    openexp._canvas.psycho.init_display(): window type = pyglet
    openexp._canvas.psycho.init_display(): waitblanking = True
    openexp._canvas.psycho.init_display(): monitor = testMonitor
    theme.py(188).load_icons: 
    Traceback (most recent call last):
      File "C:\OpenSesame 0.27\libqtopensesame\qtopensesame.py", line 1343, in run_experiment
        exp.run()
      File "C:\OpenSesame 0.27\libopensesame\experiment.py", line 295, in run
        self.init_display()
      File "C:\OpenSesame 0.27\libopensesame\experiment.py", line 595, in init_display
        canvas.init_display(self)
      File "C:\OpenSesame 0.27\openexp\canvas.py", line 54, in init_display
        exec("openexp._canvas.%s.init_display(experiment)" % experiment.canvas_backend)
      File "", line 1, in 
      File "C:\OpenSesame 0.27\openexp\_canvas\psycho.py", line 404, in init_display
        rgb=experiment.background)
      File "C:\Python27\lib\site-packages\psychopy-1.74.02-py2.7.egg\psychopy\visual.py", line 292, in __init__
        elif self.winType == "pyglet": self._setupPyglet()
      File "C:\Python27\lib\site-packages\psychopy-1.74.02-py2.7.egg\psychopy\visual.py", line 889, in _setupPyglet
        style=style
      File "C:\Python27\lib\site-packages\pyglet\window\win32\__init__.py", line 423, in __init__
        super(Win32Window, self).__init__(*args, **kwargs)
      File "C:\Python27\lib\site-packages\pyglet\window\__init__.py", line 686, in __init__
        self._create()
      File "C:\Python27\lib\site-packages\pyglet\window\win32\__init__.py", line 452, in _create
        width, height = self._client_to_window_size(self._width, self._height)
      File "C:\Python27\lib\site-packages\pyglet\window\win32\__init__.py", line 808, in _client_to_window_size
        rect.right = width
    TypeError: int expected instead of numpy.int32 instance
    
  • edited October 2012

    Thanks, again!

    My guess is, though, that (as you mentioned yourself) that most of these problems result from your configuration, which has a few older versions of the various packages installed. Perhaps you could check if the problems occur as well with the pre-release as packaged?

    And/ or you could update all the relevant packages. In the environment that I use to build the pre-release, the packages are as follows (also note NumPy, which is not listed by modules() but may be relevant).

    >>> modules()
    OpenSesame 0.27~pre16
    Python 2.7.3
    Expyriment  (Revision ; Python 2.7.3)
    PyAudio 0.2.4
    PyGame 1.9.1release
    PyOpenGL 3.0.1
    PyQt 4.9.4
    PySerial 2.6
    PsychoPy 1.74.03
    Pyglet 1.1.4
    >>> import numpy
    >>> numpy.version.version
    '1.6.2'

    Cheers!

    There's much bigger issues in the world, I know. But I first have to take care of the world I know.
    cogsci.nl/smathot

  • edited 9:01AM

    My numpy version is 1.6.1.

    I can confirm that the problem is caused by my source environment. I had another installation of Python 2.6 that I forgot about, and ran OS 0.27-pre15 with that version. The problem of the frozen mouse cursor did not occur.

    The thing is that my python 2.7 environment (which I also use for OpenSesame) was installed by Python(x,y) and I don't dare to make that many changes to it as this might damage the 'ecology' of the suite. Updating the PyQt packages already caused some weird error messages in spyder's console (although everything still worked fine). Python(x,y) (peculiarly) does not allow a separate python 2.7 installation beside its own. Luckily there will be an update of Python(x,y) soon, in which they bump all the packages to its latest versions. I will install it after its release and test again for this problem. For now, you can mark this one down as not a bug of OpenSesame (one down, a only a few to go...)

    The keyboard lockout on start-up however does occur in different installations (Ubuntu, Win 7, Win XP) so probably is not caused by my specific python environment..

    Cheers!

Sign In or Register to comment.