Re: [igraph] i386 -> x86_64 madness!!

[Marco]

Dear list,

This is what gdb spits out when i run my python script on x86_64: on
x86 works like a charm.
I have also noticed that most probably the problem's around the
bfsiter thingie: if i run a fragment of the code with ipython all
works beautifully until i call the bfsiter thing.

Response of gdb:

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 46912496292224 (LWP 27693)]
0x00002aaaaf609119 in igraphmodule_i_attribute_destroy
(graph=0x2aaaaab39e60) at src/igraphmodule.c:368
368           Py_DECREF(attrs[i]);

Here's the code i ran to obtain that SISGSEGV:
\\\\
from igraph import *

def conta_cammini(graph,distance):
    hits=0
    for v in graph.vs():
        for vertex, d, parent in graph.bfsiter(v, advanced=True):
            if d == distance:
                hits+=1
    return hits

lista=[]
for i in range(10):
    g=Graph.Erdos_Renyi(1000,2/float(999))
    lista.append(conta_cammini(g,int(sys.argv[1])))
    sys.stdout.write('.')
    sys.stdout.flush()
print '\n'
print lista
\\\

I wrote this code as a check, 'cause i had the impression that the
problem was around bfsiter. The idea is to take a vertex, and look for
paths of length 'distance' (comes from sys.argv[1]) and count them for
each vertex of the graph.

Thanks,

marco


--

è il gioco della vita,
la dobbiamo preparare
che non ci sfugga dalle dita
come la sabbia in riva al mare.

Lucio Dalla



> On Thu, Nov 6, 2008 at 12:42 PM, Tamas Nepusz <address@hidden> wrote:
>> Hi Marco,
>>
>>> Same script, just copied over a fedora core 8 x86_64 machine, it returns:
>>>
>>> python: vector.pmt:448: igraph_vector_size: Assertion `v->stor_begin
>>> != ((void *)0)' failed.
>>> Aborted
>>
>> Unfortunately I can't tell you more unless you can provide us with a
>> corresponding stack trace. Try to run your script in gdb and send us the
>> backtrace obtained at the point where the above assertion failed. It would
>> be even better if you could use a debug build of igraph as it provides the
>> exact line numbers in the stack trace.
>>
>> --
>> Tamas
>>
>>
>>
>> _______________________________________________
>> igraph-help mailing list
>> address@hidden
>> http://lists.nongnu.org/mailman/listinfo/igraph-help
>>
>