[Gnuplot-py-users] rapid multiple construction of Gnuplot.Gnuplot object and calls to .plot() produces empty graphs

[Ben W. <ben...@ot...>]

Hi

The pseudo code representing my code is as follows:

def plot(gnuCommands, data, fileID)
     #gnuCommands is an array of strings
     #data is a 2D numpy array with data[0]
     #giving the x-axis and data[1] giving the
     #y-values over the x-axis for some function
     g = Gnuplot.Gnuplot()
     fileName = getFileName(fileID)
     for command in gnuCommands:
         g(command)
     g('set terminal postscript enhance color')
     g('set output "fileName"')
     g.plot(data[0],data[1])
     g.close()

def foo(args):
     #args is an array of 2D numpy arrays
     gnuCommands = getCommands()
     for arg in args:
         plot(gnuCommands, arg, getFileID())

The result of this is that the last call to plot produces a correct 
plot. All preceding calls produce an empty plot. The only fix I've found 
is to initialise g in foo and pass it to plot. Thus I'm assuming that my 
problem is due to something like the temporary files for each g.plot() 
call being removed when a new Gnuplot.Gnuplot object is instantiated.

Am I right?

I'd like to keep the initialisation of Gnuplot.Gnuplot within the plot 
method (in my case the plot method is in a different class and I'd 
prefer not to include needless requirements for initialisation on other 
classes that use this method).

What steps can I take to fix this? E.g. if the temporary file thing is 
the problem is there a way to make them permanent? (It's not a bad thing 
in my case to have the data output in such a way).

Cheers,
Ben