I have a very strange bug. When opening figures in python, if I hover the mouse cursor above the figure toolbar, everything freezes for about a minute.
I tried:
1. Both from Spyder and using python from bash.
2. Using tkAgg as well as WX and Qt5Agg. By setting before plotting: ,e.g., matplotlib.use('Qt5Agg')
The behavior was similar in all cases.
Would appreciate any advice here. My OS is Ubuntu.
I used ipython and plot a graph and type plt.show().
This pops up a graph window. But, when I close the window,
the ipython does not go back to prompt and stays non-responding.
How can I close the graph window properly and go back to ipython ?
Thanks.
I am using ipython with python3.4 via a virtualenv and had to change matplotlib's backend in order to have the interactive plotting window show up. But when I try to plot not in the ipython shell but import a script that does the plotting for me, I can not use pyplot.close('all') (therefore multiple open windows are lingering in the back that can be open with plt.show())without getting the error:
can't invoke "event" command: application has been destroyed
while executing
"event generate $w <<ThemeChanged>>"
(procedure "ttk::ThemeChanged" line 6)
invoked from within
"ttk::ThemeChanged"
I can get rid of this behavior by explicitly using matplotlib.use('agg') but then pyplot.show() is of course not working anymore. I guess i fundamentally do not understand how to choose an appropriate backend. When I use matplotlib.get_backend() it gives me 'TKAgg'.
Is it possible to be able to close windows in a script while being able to open interactive windows or is this behavior generally not wanted?
I have a program written using PyQt and Matplotlib which I am using to display stellar spectra.
The program has a dialog box which I can use to select the required spectra and adjust the X/Y scales and other things.
When changes are made in the dialog box, I have the code do (with matplotlib.pylab imported s plt)
plt.clf()
(replot the newly-selected spectra)
plt.show()
This works fine with older versions of Matplotlib (1.0.0)
However with later versions of Matplotlib (I'm trying 1.3.1), I get strange results. I'm OK the first time round but each subsequent attempt fails to update the plot. I also get lots of messages:
QCoreApplication::exec: The event loop is already running
each time the program tries to invoke the above code and the plot is not updated.
If I change the last part of my code to say
plt.show(False)
That makes the messages go away but the plot is still not updated.
The only way I can make the plot get updated is to change the
plt.clf()
to
plt.close()
but that looks horrible as the display window closes and reopens every time I select a different spectrum.
Is there a better way of doing all this?
I'd rather use the new version of Matplotlib if I can as I also need to use the newer versions of Scipy which has routines in I am using which aren't in the older versions. At the University I'm working it's either "all old" or "all new" as far as versions of the software are concerned.
Thanks for any help you can give.
Solution by OP.
I do:
matplotlib.interactive(True)
and don't bother with any parameters to show() and it does what I want.
I'm new to PythonXY and Matplotlib. I installed PythonXY (v2.7.3.1) in default full mode.
I use the "IPython Qt Console" application.
I draw something using matplotlib.pyplot (imported as plt).
For example. When plt.plot([1,3,2,4]), the figure display in the same IPython console immediately.
If this, I cannot add some other properties, plt.title, plt.xlabel, plt.ylabel, or more. Why? And how can I draw figures in another window, adding more properties, and making the figure not display until plt.show()?
If you select from the Interactive Consoles dropdown on the Python(x,y) Home launcher, "IPython (Qt)" and then click either the Console 2 or cmd.exe button, it should then run IPython (Qt) with the Qt4Agg backend which will allow you to plot in a separate window and apply titles and so on. For more info see What is a backend.
What Python(x,y) in your example above is doing is launching IPython with the pylab inline backend which is different from the standard backends so your commands aren't having any affect, similar behaviour is noted in this issue on github. It doesn't seem possible to change the backend once IPython has been launched with inline. I'm not sure where the Python(x,y) options are set or which script is called to launch the item in the Applications dropdown.
Worth taking a look at Anaconda as a free Python distribution with a scientific focus and regular updates. Anaconda doesn't by default load pylab into IPython so you can choose the backend after launching IPython.
In the IPython command-line interface, the figure is shown and updated at each plot function.
With the inline option, Matplotlib generates an image that automatically inserted in the output line.