[Pysparse-developers] Bug from Debian user

[Adam C P. IV <haz...@de...>]

Hi,

The following segfault-inducing code came in from a user (details at
http://bugs.debian.org/535318 ):

from numpy import *
from pysparse import *
m = 4
n = 3
nodec=0
A = spmatrix.ll_mat(m*n, m*n)
H = 3
V = 2
a = [0,0,0,] #row index
b = [0,1,n] #column index
val = [H+V, -H, -V] #node value

w=m*n

nodec=0
w=m*n
bottomr=w-n
for i in range(w):
	for j in range(w):
		if (nodec)<bottomr:#check necessary to make sure we don't 
		#try puting south neighbours for the bottom row of nodes
			b.append(nodec+n)
			a.append(nodec-1)#row index, we skipped the first node
			val.append(-V)
		nodec=nodec+1
 
A.put(val,a,b)
print A

for i in range(w):
	for j in range(w):
		if (nodec-n)>=0: #check necessary to make sure we don't 
		#try puting north neighbours for the top row of nodes
			b.append(nodec-n)
			a.append(nodec-1)#row index
			val.append(-V)
        nodec=nodec+1


 
A.put(val,a,b)
print A

Also a log of crash:
In [20]: a
Out[20]: [1, 3, 5]

In [21]: b
Out[21]: [2, 3, 5]

In [22]: c
Out[22]: [4, 5, 9]

In [23]: A.take(c,a,b)
Segmentation fault

Any ideas?  Bug in user's code, pysparse, or Debian packaging?

Thanks,
-Adam
-- 
GPG fingerprint: D54D 1AEE B11C CE9B A02B  C5DD 526F 01E8 564E E4B6

Engineering consulting with open source tools
http://www.opennovation.com/