Hi Mario,
thanks a lot for your response!
1) Since we downgraded to 3.0.12 our experiments do run again.
2) The kbcheck issue(?) is present maybe already in this version, so
most likely a bug somewhere else. Further instructions are welcome!
The kernel is most recent Dec 2016 although an old version 3.13.0.
$ lsb_release -a
Description: Ubuntu 14.04.5 LTS Codename: trusty
$ uname -a
Linux xxxx 3.13.0-106-generic #153-Ubuntu SMP Tue Dec 6 15:44:32
UTC 2016 x86_64 x86_64 x86_64 GNU/Linux
$ glxgears -info
GL_RENDERER = Gallium 0.4 on llvmpipe (LLVM 3.4, 256 bits)
GL_VERSION = 2.1 Mesa 10.1.3
GL_VENDOR = VMware, Inc.
$ apt-cache policy xserver-xorg-video-nouveau
Installiert: 1:1.0.10-1ubuntu2
Installationskandidat: 1:1.0.10-1ubuntu2
Versionstabelle:
*** 1:1.0.10-1ubuntu2 0
500
http://de.archive.ubuntu.com/ubuntu/ trusty/main amd64
Matlab R2016a
>> version
9.0.0.341360 (R2016a)
>> PsychtoolboxVersion
3.0.12 - Flavor: - Corresponds to SVN Revision
For more info visit:
https://github.com/Psychtoolbox-3/Psychtoolbox-3
>> PsychtoolboxRoot
/opt/ptb-3.0.12/Psychtoolbox/
>> KbStrokeWait(-1)
y # <-- this is where on an office desktop
a # control gets immediately back to
y # Matlab main window: >> y
^C
Operation terminated by user during KbCheck (line 229)
In KbWait (line 143)
[isDown, secs, keyCode, deltaSecs] = KbCheck(deviceNumber);
In KbWait (line 132)
[secs, keyCode, deltaSecs] = KbWait(deviceNumber, 0, untilTime);
In KbStrokeWait (line 32)
[secs, keyCode, deltaSecs] = KbWait(deviceNumber, 3, varargin{:});
>> [keyboardIndices, productNames, allInfos]= GetKeyboardIndices
keyboardIndices =
3 4 5
productNames =
'Virtual core XTEST keyboard' 'Power Button' 'Sleep Button'
allInfos =
[1x1 struct] [1x1 struct] [1x1 struct]
Only one keyboard (bluetooth) and mouse (same bluetooth) and
additionally the ResponsePixx button-box (uninitialized at that time).
3) Of course you are welcome at NIP any time! I'm looking forward ...
See you,
Uli
On 01/30/2017 01:31 AM, mario.kleiner@... [PSYCHTOOLBOX] wrote:
> XX---In PSYCHTOOLBOX@yahoogroups.com, <h_schuett1990@...> wrote :
>
> Dear all,
> -> Hi Heiko
>
> we have a problem with a new install of psychtoolbox (3.0.14). Psychtoolbox now requires us to varify that the setup works using the test BitsPlusIdentityClutTest(1, 1). This test works nicely, everything is displayed correctly and after pressing 's' the command window claims saving the config worked.
>
> -> Saving is not really needed if it works at default settings, as it always should on AMD gpu's under Linux. Doesn't hurt either though afair.
>
> After pressing ESC we cannot answer the question whether everything is fine as all keystrokes just print into the command window and return/enter is not registered. Thus the only way to end the function is to press CTRL-C which ends the program, but seems not to yield a vaildation.
>
> What could be the problem here?
>
> -> Hmm. I run that test probably multiple times each month on Linux machines as part of my standard test procedures, also on KUbuntu, and the code doesn't do anything special, so this is surprising. However, i almost always run that test like almost all of my testing on Octave and not Matlab so i don't have to deal with all the annoying license manager and registration hassle. I'd have to test what happens on Matlab. All the keyboard input runs through KbCheck, the only difference is that in the scripts main while loop, KbCheck is used to only poll the default keyboard instead of KbCheck(-1), whereas the GetEchoString() command used to ask the y/n question internally uses KbCheck(-1) to poll all keyboards. Not sure why that difference could cause trouble, but do you have multiple keyboards connected?
>
> What happens if you call KbStrokeWait(-1) on the command window? Does it respond to a keypress or does it hang?
> What is the full detailed output of [keyboardIndices, productNames, allInfos]= GetKeyboardIndices
>
> -> As a workaround for the moment you can just use a copy of that script, hacked to just skip that query.
>
> Our setup is a runs a Kubuntu with a 3.13 Kernel with display on a VIewpixx lcd monitor with Datapixx data collection. Additional Information can be found in the attached text-file, which lists the output during the run of BitsPlusIdentityClutTest(1, 1).
>
> -> i wonder if you are hitting some bug in your X-Server, or some interaction bug between the old X-Server and current PTB? Your KUbuntu system seems to still run the original Linux kernel and display stack of the April 2014 14.04.0 LTS release. Haven't tested current PTB for compatibility in over a year, only the upgraded kernel/display hardware enablement stack of 14.05.5-LTS still gets tested.
>
> Which reminds me i still haven't visited Felix "new" lab in all these years. It's ridiculous how bad i am at organizing these things.
>
> -mario
>
> best wishes
> Heiko Schütt
> (Wichmann-lab)