Re: [Plplot-general] wingcc Freeze

[John D. <jb_...@bt...>]

Hi Alan

Thanks for your comprehensive reply, did you see my second posting regarding the freeze occurring when window events occur? I'm tending to think it is a window event issue in the wingcc code. I did monitor window resources during my program run and freeze, memory usage didn't appear to be a problem.

I will digest your email and do some further diagnosis as you suggest.

I'm using Windows only because I have to at the moment. I intend to migrate to a Unix platform, which I am more familiar with, as soon as possible.

Kind regards

John

Sent from my iPhone

> On 14 Mar 2014, at 21:33, "Alan W. Irwin" <ir...@be...> wrote:
> 
>> On 2014-03-14 20:21-0000 John Duffy wrote:
>> 
>> Hi
>> 
>> I have the following pseudo code which work well until it freezes...
>> 
>> plsdev("wingcc");
>> plspage(0, 0, 1600, 800, 0, 0);
>> plspause(0);
>> plinit();
>> 
>> for (...) {
>> plenv(); // set new x & y max and min parameters
>> plline(); // data to plot
>> }
>> 
>> plend();
>> 
>> Like I say, it works well until it freezes after a few hundred iterations of displaying data. I would appreciate any help in pointers where to look for problems, my code or within the "wingcc" driver. When the code freezes, Windows reports the plotting window as not responding.
> 
> Hi John:
> 
> My guess is you are encountering some sort of memory management issue
> (heap corruption) or memory leak in the wingcc code or in the PLplot
> core library that is triggered by all your iterations.
> 
> My bet is on a memory management issue for the combination of
> wingcc and the general PLplot -np option that you are using. My
> impression is that combination does not work as well as it should. For
> example, although results for the wingcc device normally display well
> for me on the Wine platform, they only show "transparent" results
> (just the frame for the window with everything else leaking through
> from the display below) if the -np PLplot option is used.  Of course,
> this strange behaviour might just be the result of a Wine bug, but the
> other alternative is some memory management issue with how -np is
> implemented for the wingcc device driver that is creating a wide
> variety of peculiar results on various platforms.
> 
> To distinguish whether the problem is in wingcc + the -np option or in
> the PLplot core, what happens if you try the equivalent for the svg
> device?  (The -np option is ignored in that case since svg is a
> file-oriented device.  However, use the command line options
> 
> -fam -fflen 3
> 
> so that each of your plot pages [generated by plenv] will be written
> to a separate file rather than just the first page written to one file
> with the rest of the pages dropped).  I also suggest you keep the
> number of iterations small until you get an idea of the disk space
> consumed by all the different svg page files you are going to generate
> by a large number of iterations.
> 
> If you do have any trouble with the svg case could you simplify the
> example as much as possible while still demonstrating the issue, and
> then send me that simplified code so I can try it for myself? I would
> be happy to run it here on Linux under valgrind to check for any
> memory management or memory leak issues.  Unfortunately, I cannot do
> such a check for wingcc since wingcc is Windows only and valgrind is
> Unix only.
> 
> Alan
> __________________________
> Alan W. Irwin
> 
> Astronomical research affiliation with Department of Physics and Astronomy,
> University of Victoria (astrowww.phys.uvic.ca).
> 
> Programming affiliations with the FreeEOS equation-of-state
> implementation for stellar interiors (freeeos.sf.net); the Time
> Ephemerides project (timeephem.sf.net); PLplot scientific plotting
> software package (plplot.sf.net); the libLASi project
> (unifont.org/lasi); the Loads of Linux Links project (loll.sf.net);
> and the Linux Brochure Project (lbproject.sf.net).
> __________________________
> 
> Linux-powered Science
> __________________________