(09/02/2013) MatPlotLib 1.2.0 is now released for MeeGo Harmattan, just 1 month after the upstream release (08/01/2013), with interactive graphical backend on Nokia N9/N950 and many new features.

MatPlotLib 1.2.0 Harmattan is based on Ubuntu 13.04 version and MatPlotLib 1.2.0 source code, without Python 3 support, no debug package, adapted dependencies to MeeGo Harmattan, etc.

See 1st post por details about installing and using MatPlotLib 1.2.0.

Time to load MatPlotLib 1.2.0 on Nokia N9 :
- 0.7-0.9 s via "from matplotlib import *" ;
- 2.0-2.3 s (with NumPy 1.4.1 or 1.7.0) via "import pylab" or "from pylab import *".
Comparing with MatPlotLib 1.0.0 :
- 0.7-0.9 s via "from matplotlib import *" ;
- 1.5-1.7 s (with NumPy 1.4.1 or 1.7.0) via "import pylab" or "from pylab import *".

Benchmarks of MatPlotLib examples (see posts #1 and #4) with NumPy 1.4.1 (from Nokia repositories) on Nokia N9 :
simple_plot.py : 2.9s (1.0.0), 4.4s (1.2.0)
surface3d_demo.py : 10.0s (1.0.0), 9.7s (1.2.0)
convfield.py : 2.8s (1.0.0), 4.6s (1.2.0)
plotfx.py : 6.1s (1.0.0), 8.4s (1.2.0)
contourfxy.py : 7.4s (1.0.0), 10.2s (1.2.0)
densityfxy.py : 7.5s (1.0.0), 10.3s (1.2.0)
vectorfieldFxy.py : 5.8s (1.0.0), 8.2s (1.2.0)
surfacefxy.py : 19.7s (1.0.0), 16.6s (1.2.0)

The same, but using new NumPy 1.7.0 on Nokia N9 :
simple_plot.py : 3.2s (1.0.0), 4.8s (1.2.0)
surface3d_demo.py : 12.1s (1.0.0), 10.9s (1.2.0)
convfield.py : 3.2s (1.0.0), 4.8s (1.2.0)
plotfx.py : 6.4s (1.0.0), 8.9s (1.2.0)
contourfxy.py : 8.3s (1.0.0), 10.9s (1.2.0)
densityfxy.py : 8.0s (1.0.0), 10.6s (1.2.0)
vectorfieldFxy.py : 6.5s (1.0.0), 8.6s (1.2.0)
surfacefxy.py : 22.8s (1.0.0), 19.0s (1.2.0)

As we see above :
- MatPlotLib 1.2.0 is slower than v1.0.0, except for surface 3D plots;
- using NumPy 1.7.0 is slower than v1.4.1.