The case of the disappearing windows.
- To: mathgroup at yoda.physics.unc.edu
- Subject: The case of the disappearing windows.
- From: mcwilljg at euler.sfasu.edu
- Date: Mon, 5 Oct 92 13:58:33 CDT
I have written a number of Mma procedural programs for use in a numerical analysis class that I teach. These programs are files containing Mma commands and are executed using the <<filename command. The students must enter a function (which is then parsed), and constant values specific to the particular procedure. They are given an option to plot the entered function. Since they are usually running the kernel remotely, they can choose the type of output device that they need(terminal, tektronics or Mma window). My problem is this: In any single session, if the programs are run in a Mma window, after a number of executions(not always the same number), Mma dies immediately after one of the input windows(not always the same one) disappears after accepting input. This is not a logic problem, the programs run on any legitimate example, nor is it a case of bad input values, the same example will run 4 or 5 times and MMa will die on the 6th time. Furthermore, if the programs are run remotely, no problems are experienced at all. My suspicion is that perhaps some variables or functions must be cleared after each run or some sort of initialization or resetting must be done after each run in order to avoid a buffer overflow when executing in a window. NOTE: SWAPFILE does _not_ increase is size after a MMa window dies in this manner. Any suggestions??