From: Sean M. <sma...@gm...> - 2014-09-09 01:27:07
|
The problem is still there if I put in a long call (5 - 10 seconds) to time.sleep after the plot. I am running this within a python interpreter, so the python thread gets to run as long as it needs. Just to mess around a bit more, I also tried it by putting it within a file then running it from the shell, with something like $ python2 analysis.py And I'm still getting the same factorize error. On Mon, Sep 8, 2014 at 7:47 AM, Michael Haggerty <mh...@al...> wrote: > On 09/08/2014 02:38 AM, Sean Mattingly wrote: > > I would just like to report my experience / trouble so that others may > > find solace if they have a similar problem. > > > > I have code which makes a basic 2 dimensional Gaussian function. I want > > to make a basic surface plot of this function, just to see what it looks > > like. I have found that using Gnuplot.GridData function with the > > optional filename argument enabled, so that it creates a permanent file, > > works. If I leave out the "filename=test.bin" option, the program > > defaults to the temporary FIFO class, and Gnuplot (4.6) is unable to > > read and properly facorize the matrix. > > Problems like this could occur if the FIFO gets deleted too soon or the > Python thread that is supposed to write into the FIFO is killed too > soon. Does the problem go away if you put a long call to time.sleep > after the plot command? > > It is also possible that newer versions of Gnuplot want to read through > the file twice. This would be incompatible with the FIFO code, which is > only prepared to write the content into the FIFO a single time. > > Michael > > -- > Michael Haggerty > mh...@al... > > |