matplotlib-users Mailing List for matplotlib

The pick radius for the X and Y axis artists are one-sided; is that
intentional?

Original (in class XAxis), file axis.py:

    def contains(self,mouseevent):
        """Test whether the mouse event occured in the x axis.
        """
        if callable(self._contains): return self._contains(self,mouseevent)

        x,y = mouseevent.x,mouseevent.y
        try:
            trans = self.axes.transAxes.inverted()
            xaxes,yaxes = trans.transform_point((x,y))
        except ValueError:
            return False, {}
        l,b = self.axes.transAxes.transform_point((0,0))
        r,t = self.axes.transAxes.transform_point((1,1))
        inaxis = xaxes>=0 and xaxes<=1 and (
                   (y<b and y>b-self.pickradius) or
                   (y>t and y<t+self.pickradius))
        return inaxis, {}


Corrected:

    def contains(self,mouseevent):
        """Test whether the mouse event occured in the x axis.
        """
        if callable(self._contains): return self._contains(self,mouseevent)

        x,y = mouseevent.x,mouseevent.y
        try:
            trans = self.axes.transAxes.inverted()
            xaxes,yaxes = trans.transform_point((x,y))
        except ValueError:
            return False, {}
        l,b = self.axes.transAxes.transform_point((0,0))
        r,t = self.axes.transAxes.transform_point((1,1))
        inaxis = xaxes>=0 and xaxes<=1 and (
                   (y<b+self.pickradius and y>b-self.pickradius) or   # <---
change on this line
                   (y>t-self.pickradius and y<t+self.pickradius))   # <---
change on this line
        return inaxis, {}

-- 
Daniel Hyams
dh...@gm...

Hi Stoyan,

I had this problem, refer to
http://sourceforge.net/mailarchive/forum.php?thread_name=1315605880.26850.7.camel%40GARDEN&forum_name=matplotlib-users

It was dismissed as a code error (which I would perhaps dispute).

You have two options:-

- Adopt the approach seen here:
http://matplotlib.sourceforge.net/examples/user_interfaces/embedding_in_qt4.html

- Or use the work around:
import matplotlib
# Free memory
matplotlib.pyplot.close()

Regards
Matt Earnshaw

On Sat, Oct 8, 2011 at 9:22 PM, <sto...@ma...> wrote:
> ------------------------------
> Message: 7
> Date: Sat, 08 Oct 2011 21:22:45 +0400
> From: ????? ???????<sto...@ma...>
> Subject: [Matplotlib-users] Memory leak - canvas.draw() + pyqt4
> To:mat...@li...
> Message-ID:<E1R...@f2...>
> Content-Type: text/plain; charset="utf-8"
>
> Memory leak - canvas.draw() + pyqt4
> Hi, my name is Stoyan.
> ??? I use the following packages: python 2.6 , PyQt 4.8.5-1, numpy 1.6 and matplotlib 1.0.1. In my program Matplotlib library I have used a GUI application created by "PyQt". In the program there is a loop that updates a graph by calling:
> ?self.fig.canvas.draw()
> ?? But every time the graph is updated, the memory usage goes up.I the address listed bellow there is an explanation of the reason for the issue however, even though
> https://github.com/matplotlib/matplotlib/pull/89?
> ?? I have followed all instructions that have been specified in link,there is still a flaw in the programme due to a memory leak, though not a big one.?? I wondere whether there is alternative way to discharge this leak of memory and if so could you assist in solving it.

Memory leak - canvas.draw() + pyqt4 
Hi, my name is Stoyan.
    I use the following packages: python 2.6 , PyQt 4.8.5-1, numpy 1.6 and matplotlib 1.0.1. In my program Matplotlib library I have used a GUI application created by "PyQt". In the program there is a loop that updates a graph by calling: 
 self.fig.canvas.draw()
   But every time the graph is updated, the memory usage goes up.I the address listed bellow there is an explanation of the reason for the issue however, even though
https://github.com/matplotlib/matplotlib/pull/89 
   I have followed all instructions that have been specified in link,there is still a flaw in the programme due to a memory leak, though not a big one.   I wondere whether there is alternative way to discharge this leak of memory and if so could you assist in solving it. 

s kemp <6b6...@gm...> writes:

> The output i get from this code is two identical pages of PDF where
> the x limits are set to 3,5. I understand this is because I am simply
> making two identical references to the figure object in the list and
> by modifying fig1 i also affect the fig1 that is 'stored' in the first
> position of the array. My question is how do I avoid this? 

My recommendation is to construct multiple figure objects. Factor the
shared code into a function so it is easier to modify just what you want
to be different between the figures:

------------------------------------------------------------------------
import matplotlib.figure as fig
from matplotlib.backends.backend_pdf import PdfPages, FigureCanvasPdf

def plot_data(xdata, ydata):
    """Returns pair of figure and axis"""
    f = fig.Figure(figsize=(2,2))
    ax = f.add_subplot(111)
    ax.plot(xdata, ydata)
    return f, ax

xdata = [1,2,3,4,5]
ydata = [5,4,3,2,1]

fig1, ax1 = plot_data(xdata, ydata)
fig2, ax2 = plot_data(xdata, ydata)

figures = []

ax1.set_xlim([0,3])
figures.append(fig1)

ax2.set_xlim([3,5])
figures.append(fig2)

pdf = PdfPages('output.pdf')

for figure in figures:
       canvas = FigureCanvasPdf(figure)
       figure.savefig(pdf, format="pdf")

pdf.close()
------------------------------------------------------------------------

-- 
Jouni K. Seppänen
http://www.iki.fi/jks