Re: [Plplot-general] Resizing issues of plplot window using python and qt (segfaults)

[Alan W. I. <ir...@be...>]

On 2015-03-16 12:41+0100 Peter Holtermann wrote:

> Dear all,
>
>
> I have recently started to try plplot to plot realtime data. I have
> extensively used matplotlib but found it for realtime much to slow. So I
> tried plplot and I got pretty fast nice results using plplot.
> I have anyhow a segfault problem with my python code and I do not fully
> understand why.
>
>
> What I experienced when e.g. applying the examples is that a resizing of
> the window does resizes the axes only when the window is resized
> according to the axes aspect ratio. So a resizing of the window in e.g.
> only the x-direction does not resize the plplot axes.
>
>
> What I found to work is that after a window resize you cleanup your
> existing plplot widgets ( calling plend(), plfreeqtdev() ) and start all
> over. This works quite well, but I get random segfaults. Attached is my
> python code, it should work out of the box.
>
>
> Any help is appreciated!

Hi Peter:

Thanks for your interest in PLplot.  I am glad to hear it is faster
than matplotlib, but sorry to hear that you are running into segfaults
for your test of our pyqt4 binding.

We take memory management errors (which often produce segfaults)
seriously.  However, it may be a long process to sort out this issue,
but to help get that debug process started, I and others do test our
pyqt4 binding via examples/python/pyqt4_example.py (e.g., by using the
cmake option -DBUILD_TEST=ON, and after cmake is completed running

make test_pyqt4_example

in the build tree or as part of some larger test such as

make test_interactive

in the build tree.

We run such tests routinely as part of due diligence leading up to
releases, and so far nobody has reported segfaults with that example.
Of course, that is a "for what it is worth" comment because memory
management errors sometimes don't lead to segfaults or any other
obvious symptom, and also because pyqt4_example.py is a simpler test
for our pyqt4 binding than your test case.  Nevetheless, you should do
that same test yourself just to make sure you can replicate our good
test result for this case.

Your post caught my eye because you mentioned resizing.  That
typically uses plbuf capabilities, and a number of bug fixes have just
been completed for plbuf.  So just now for the git master tip version
of PLplot I ran the test_pyqt4_example target above and did a lot of
resizing of the resulting GUI without segfault (or other) issues, and
you might want to try the same thing.

Also, I suggest you try that test and also your own example using the
git master tip version of PLplot just in case the recent plbuf fixes
have solved the segfault problem for you.

And more about that master tip version in my next post.

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
__________________________