Howdy, are there fundamental reasons to keep this code around in backends/__init__.py:use() ? if 'matplotlib.backends' in sys.modules: if warn: warnings.warn(_use_error_msg) return I am now testing the new IPython gui stuff, and if I comment that out and call pyplot.switch_backend() when I load mpl (along with the qt patch I sent earlier), I can hop around backends, at least with Qt4 and Wx: In [10]: %pylab qt Welcome to pylab, a matplotlib-based Python environment. Backend in use: Qt4Agg For more information, type 'help(pylab)'. In [11]: run simpleplot.py In [13]: %pylab wx Welcome to pylab, a matplotlib-based Python environment. Backend in use: WXAgg For more information, type 'help(pylab)'. In [16]: run simpleplot.py In [17]: figure() Out[17]: <matplotlib.figure.Figure object at 0xa311b0c> In [18]: plot(sin(linspace(0,2*pi,200)**2)) Out[18]: [<matplotlib.lines.Line2D object at 0xa49862c>] In [19]: %pylab qt Welcome to pylab, a matplotlib-based Python environment. Backend in use: Qt4Agg For more information, type 'help(pylab)'. In [20]: run simpleplot.py In [21]: figure() Out[21]: <matplotlib.figure.Figure object at 0xa64fb2c> In [22]: plot(sin(linspace(0,2*pi,200))) Out[22]: [<matplotlib.lines.Line2D object at 0xa67d26c>] etc... I see lockups trying to throw Tk or Gtk in the mix, but being able to switch from Wx to Qt and back would be nice, especially since it would allow using Mayavi and other enthought apps together with Qt-based tools. mpl closes open windows on backend switch, but at least you can go back and forth, which is better than nothing and can be useful for debugging. This is still very early/experimental ipython code, I just want to start getting a feel for the mpl-side of the equation... Cheers, f