I too have run across the infamous pyglet memory leak, that occurs when one needs to update text components at the frame level (with a text variable typically computed in a separate code component, also at frame level).
The leak/bug crashes a running script with an error message of the sort:
File "C:\Program Files (x86)\PsychoPy2\lib\site-packages\psychopy-1.83.04-py2.7.egg\psychopy\visual\window.py", line 541, in flip
thisStim.draw()
File "C:\Program Files (x86)\PsychoPy2\lib\site-packages\psychopy-1.83.04-py2.7.egg\psychopy\visual\text.py", line 603, in draw
self._pygletTextObj.draw()
File "C:\Program Files (x86)\PsychoPy2\lib\site-packages\pyglet\font\__init__.py", line 565, in draw
self._layout.draw()
File "C:\Program Files (x86)\PsychoPy2\lib\site-packages\pyglet\text\layout.py", line 852, in draw
self.batch.draw()
File "C:\Program Files (x86)\PsychoPy2\lib\site-packages\pyglet\graphics\__init__.py", line 544, in draw
func()
File "C:\Program Files (x86)\PsychoPy2\lib\site-packages\pyglet\graphics\__init__.py", line 476, in <lambda>
(lambda d, m: lambda: d.draw(m))(domain, mode))
File "C:\Program Files (x86)\PsychoPy2\lib\site-packages\pyglet\graphics\vertexdomain.py", line 313, in draw
glDrawArrays(mode, starts[0], sizes[0])
WindowsError: exception: access violation reading 0x00000010
Exception TypeError: "'NoneType' object is not callable" in <bound method Server.__del__ of <pyolib.server.Server object at 0x0EFB3670>> ignored
I know the PsychoPy community has linked this to a memory leak in pyglet that was meant to have been fixed around April 2016 according to this thread. I downloaded the latest version of Pyglet (which is v1.2.4 - released apparently in Sep 2015, so before the supposed fix of the leak) and replaced its contents under c:\Program Files (x86)\PsychoPy2\Lib\site-packages\pyglet\. However, this did not fix my problem.
Unfortunately, my experiment does not make sense if I give up the frame-wise computation that feeds the text component.
An older thread that I found suggested the following workaround: in the code component that computes frame-wise the value of the text variable being displayed, only assign this variable a new value if that new value actually differs from the previous value. However, even after changing my code component this way, the pyglet error still pops up.
Assuming the bug was fixed in Pyglet, is it not v1.2.4 that reflects this fix, and should it not be updated by overwriting the contents of c:\Program Files (x86)\PsychoPy2\Lib\site-packages\pyglet\ (maintaining the original folder structure)?
Assuming the bug was in fact NOT fixed in Pyglet, can anyone suggested a different workaround that I can try to make my PsychoPy expt not crash?
Many thanks indeed for your help.
(Apologies for the double post both here and on the psychopy group; will update both threads as soon as there is a response on the other, to avoid people wasting time with double answers).
One workaround is to use TextBox instead of TextStim. To do that, one first defines a stimulus object in a code component (under "begin routine"), such as
textbox = visual.TextBox(window=win,
text='Something',
font_size=21,
font_color=[-1,-1,1],
size=(1.9, .3),
pos=(0.0, 0.25),
grid_horz_justification='center',
units='norm')
then, perhaps under the "every frame" tab, one uses code to customise & call upon the thus-defined textbox, with syntax as found on http://www.psychopy.org/api/visual/textbox.html
Related
I have a client who has asked me to fix a problem with a corrupt user.config file. The config file has enough of an XLM header for the program to load the settings, but the rest of the config file is nulls (0x00) so the config loads, but then an error occurs when attempting to access any of its member. Note: the 0x00 value pops up in the error message later.
After much research, the common suggested solution seems to be delete the user.config file and then do a My.Settings.Reset but this doesn't work for me. The only solution that seems to work is to delete the user.config file, and then shut down and restart the program. The program starts and runs fine after deleting the user.config file. It creates a new user.config with default values and goes on its merry way with no problems.
What I have tried:
First:
Dim innerExcept As ConfigurationErrorsException = CType(cex.InnerException, ConfigurationErrorsException)
My.Computer.FileSystem.DeleteFile(innerExcept.Filename)
Then various combinations of the following:
My.Settings.Reset()
My.Settings.Reload()
My.Settings.Upgrade()
No matter what I try, I get the following error. Any thoughts?
System.Configuration.ConfigurationErrorsException HResult=0x80131902
Message=Configuration system failed to initialize
[truncated for brevity]
Inner Exception 1: ConfigurationErrorsException: '.', hexadecimal
value 0x00, is an invalid character.
I'm writing a startup script that starts and logs in on all my applications I use daily. For this I'm using the PyAutoGui module, python 3, and another application that manages the layout of my windows on my 2 monitors.
def smv(username, hotkey):
# starting application
os.system("smv.exe")
# it autofocuses on the username field
pyautogui.typewrite(username)
# (passwords are the same)
pyautogui.typewrite("PASSWORD\n")
# move to the window header (where title, minimize, close, etc..)
pyautogui.moveTo(100, 10, duration=0)
# drag the window to the far right of the primary monitor (since pyautogui doesn't support multiple monitors)
pyautogui.dragTo(1910, 20, duration=1, button="left")
# press the hotkey that belongs to the window layout manager
pyautogui.hotkey("ctrl", "alt", hotkey)
smv("username", "num7")
smv("username2", "num1")
The weird thing is that the first smv() runs perfectly fine. Application starts up, logs in, moves the window and adjusts the position/size with the hotkey.
The problem occurs when the second smv() runs. Application starts up, logs in, moves to window header, but then crashes with:
Traceback (most recent call last):
File "main.py", line 15, in <module>
smv("username2", "num1")
File "D:\files\PyCharm\startup\smv2.py", line 10, in start
pyautogui.dragTo(1919, 10, duration=1, button="left")
File "d:\PortableApps\Miniconda3\lib\site-packages\PyAutoGUI-0.9.33-py3.5.egg\pyautogui\__init__.py", line 683, in dragTo
File "d:\PortableApps\Miniconda3\lib\site-packages\PyAutoGUI-0.9.33-py3.5.egg\pyautogui\__init__.py", line 274, in mouseDown
File "d:\PortableApps\Miniconda3\lib\site-packages\PyAutoGUI-0.9.33-py3.5.egg\pyautogui\_pyautogui_win.py", line 393, in _mouseDown
File "d:\PortableApps\Miniconda3\lib\site-packages\PyAutoGUI-0.9.33-py3.5.egg\pyautogui\_pyautogui_win.py", line 480, in _sendMouseEvent
PermissionError: [WinError 5] Access denied.
I'm cracking my brain trying to figure out why it crashes at the second run.
I've tried using try and except blocks (no use), I've switched the order in which the application starts (num1 before num7 and vice versa), but no use (1st one runs, 2nd crashes).
I've even tried to split them up into 2 files and running it from a third "main.py" file. Its not efficient, I know, but wanted to see if it made any difference.. you guessed it: it doesn't. It still crashes on PermissionError.
In 0.9.34, this has been fixed (or at least, the exceptions are now suppressed since the clicks still seem to work anyway). So the solution is to upgrade PyAutoGUI.
When I attempt to compile a LESS template in Visual Studio using Web Essentials, I receive an error that says "Unexpected token u" with no file name, no line number, and no column number. Why is this happening?
Go to %USERPROFILE%\AppData\Local\Microsoft\VisualStudio\12.0\Extensions which is the folder where per-user Visual Studio extensions reside. WebEssentials will be located in a subfolder with a randomly generated name.
From inside the WebEssentials folder, open up the file Resources\nodejs\tools\server\services\srv-less.js and go to line 65, which reads:
map = JSON.parse(output.map);
The problem is source map output may be the undefined value. JSON.parse can only parse strings, so it casts that to the string value "undefined" before parsing, but JSON does not recognize that as valid token. (It only understands the null value, not the undefined value.)
So... change line 65 to read:
map = JSON.parse(output.map || "null");
And voilà; LESS compilation on files with empty output works again.
Source:
https://github.com/madskristensen/WebEssentials2013/issues/1696
From my experience, this error occurs when LESS attempts to output a CSS file from a LESS file, and the resulting CSS file is empty. In my case, this happened after removing some font-face declarations, which left the resulting CSS file empty. LESS would not compile until I added a class that would output to the CSS file.
Details may be found here: https://github.com/madskristensen/WebEssentials2013/issues/1696
I'm adding this to StackOverflow because I'm unable to access Github at my workplace. I hope this helps someone.
You can also add in your less file an important comment /**/ or #charset "utf-8"; as described here https://github.com/madskristensen/WebEssentials2013/issues/1696
I just downloaded psychopy this morning and have spent the day trying to figure out how to work with builder view. I watched the youtube video "Build your first PsychoPy experiment (Stroop task)" by Jon Pierce. In his video he was explaining how to make a conditions file with excel that would be used in his experiment. I wanted to make a very similar test where images would appear and subjects would be required to give a yes or no answer to them (the correct answer is already predefined). In his conditions file he had the columns 'word' 'colour' and 'corrANS'. I was wondering if instead of a 'word' column, I can have an 'image' column. In this column I would like to upload all my images to them in the same way I would words, and have them correlated to a correct answer of either 'yes' or 'no'. We tried doing this and uploaded images to the conditions file, but we haven't had any success in running the test successfully and were hoping somebody could help us.
Thank you in advance.
P.S. we are not familiar with python, or code in general, so we were hoping to get this running using the builder view.
EDIT: Here is the error message we are receiving when running the program
#### Running: C:\Users\mr00004\Desktop\New folder\1_lastrun.py
4.8397 ERROR Couldn't find image file 'C:/Users/mr00004/Desktop/New folder/PPT Retention 1/ Slide102.JPG'; check path?
Traceback (most recent call last):
File "C:\Users\mr00004\Desktop\New folder\1_lastrun.py", line 174, in
image.setImage(images)
File "C:\Program Files (x86)\PsychoPy2\lib\site-packages\psychopy-1.80.03-py2.7.egg\psychopy\visual\image.py", line 271, in setImage
maskParams=self.maskParams, forcePOW2=False)
File "C:\Program Files (x86)\PsychoPy2\lib\site-packages\psychopy-1.80.03-py2.7.egg\psychopy\visual\basevisual.py", line 652, in createTexture
% (tex, os.path.abspath(tex))#ensure we quit
OSError: Couldn't find image file 'C:/Users/mr00004/Desktop/New folder/PPT Retention 1/ Slide102.JPG'; check path? (tried: C:\Users\mr00004\Desktop\New folder\PPT Retention 1\ Slide102.JPG)
Yes, certainly, that is exactly how PsychoPy is designed to work. Simply place the image names in a column in your conditions file. You can then use the name of that column in the Builder Image component's "Image" field. The appropriate image file for a given trial will be selected.
It is difficult to help you further, though, as you haven't specified what went wrong. "we haven't had any success" doesn't give us much to go on.
Common problems:
(1) Make sure you use full filenames, including extensions (.jpg, .png, etc). These aren't always visible in Windows at least I think, but they are needed by Python.
(2) Have the images in the right place. If you just use a bare filename (e.g. image01.jpg), then PsychoPy will expect that the file is in the same directory as your Builder .psyexp file. If you want to tidy the images away, you could put them in a subfolder. If so, you need to specify a relative path along with the filename (e.g. images/image01.jpg).
(3) Avoid full paths (starting at the root level of your disk): they are prone to errors, and stop the experiment being portable to different locations or computers.
(4) Regardless of platform, use forward slashes (/) not backslashes (\) in your paths.
make a new folder in H drive and fill in the column of image in psychopy as e.g. 'H:\psych\cat.jpg' it works for me
I just wrote a java class to generate meshes from a cylinder list stored to a ply file. I tested the files with a hand generated list of 3 cylinders. The resulting file I can open both in Meshlab and Cloudcompare.
When I use the class in my real program I have to write a mesh for more than 13000 cylinders.
Cloudcompare gives me the following error : Reading error(no access right?)
Meshlab this one : error details, unexptected eof
I already checked if my ply file contains the exact number of vertices and faces defined in the header. I also assured, there are no nan (checked for 'n','a', etc in winedit) values contained.
I can reproduce the errors with my test file from the 3 hand made cylinder file by deleting the last line. But as mentioned earlier, I already checked if the line numbers are correct (might be an empty line not caught by my eyes though, as scrolling down half a million lines is impossible).
So are there any programs available to parse the ply file for errors? Open source tools would be appreciated here. Or are the files just to large? 436302 lines to be exact. I use ascii version of ply.
Found a non open source tool called nugraf, which provides information about the corrupted line numbers.
Java seems to print NAN with '?'. For this char i did not check, so problem seems to be solved and I can debug my java software now again.