I see a significant difference in plots generated with two recent versions of matplotlib, and unfortunately not in favour of the newer one, otherwise there would be no reason to write here, I guess.
The problem is following, I plot some data wich might be undefined in some cases. For example, I plot two lines and one of them has values 'None' only. Using the 0.98.3 version there is no problem, the graph with 'None'-s is just not shown, but with 0.98.5 the y-scale is something close to zero only, if the first line is 'None'-s and I do not see the plot for other line as well.
I add simple code, which works fine with 0.98.3 and not with 0.98.5. It is definitely not difficult to make a workaround, but for some reason I like the behaviour of previous version better and if there is no need to change it, maybe we could turn it back into situation, where one line of missing data would not destroy the whole graph.
Univ. of Tartu, Estonia
import sys #sys.path.insert(1,'/home/andres/SKRIPTID/lib/python')
import numpy as np import matplotlib matplotlib.use( 'Agg' ) import pylab strm=np.array([[None,10,None,5,1],