Re: [Visualpython-users] segfault with visual 5.11 / python 2.5.4 / Debian i386

[Bruce S. <Bru...@nc...>]

I'll comment that one just gets a standard error message, not a 
segfault, on Ubuntu, Windows, and Mac.

Bruce Sherwood

Sanjoy Mahajan wrote:
> On Debian GNU/Linux (i386, 'unstable' distribution), I get a segfault
> with the following vpython script.  I'm using v5.11 as packaged in
> Debian, with Python 2.5.4 from Debian.
>
> Does anyone see the same problem?
>
> I eventually figured out one problem with the script: the "F -=
> ball.m*g" line needs a vector on the right-hand side.  But I don't think
> it should segfault even with that mistake.
>
> By the way, valgrind is not very happy with the script, and also gives
> something of a backtrace for the segfault.  I've included the end of the
> output of "valgrind --leak-check=full python /tmp/script.py"
>
> -Sanjoy
>
>
> from visual import *
>
> cd = 0.5                        # drag coefficient for a sphere
> ball = sphere(pos=(0,0,0),      # in m
>               color=color.red,
>               radius=0.05)      # in m
> ball.v = vector(10,0,10)                                 # in m/s
> ball.m = 0.5
> g = 10
> endpoint = sphere(pos=(20,0,0))
> startpoint = sphere(pos=(0,0,0))
> scene.autoscale = 0
>
> trail = curve()
> dt = 0.0001
> area = pi * ball.radius**2      # cross-sectional area of ball
> while 1:
>     rate(int(1/dt))
>     ball.pos += ball.v*dt # use updated velocity to update position
>     trail.append(pos=ball.pos)
>     F =  -0.5 * cd * area * ball.v * ball.v.mag
>     F -= ball.m*g
>     ball.v += F/ball.m*dt            # force changes the velocity
>
>
> And here is the valgrind output:
>
>
>
>