Closed 85eec1a4-3d04-4b4d-b711-d4db03337c41 closed 16 years ago
In state 6 and state 16, the two lines
if hb > nu.k: # update hb since we are backtracking (not in [1])
hb = nu.k # recall hb is the longest common ancestor of rho and nu
need to be wrapped with an if lab:
clause. Will post a patch once my build finishes.
Attachment: trac3607.patch.gz
Nice job rlm, this nails the issue and valgrind now gives graph_isom.py a clean bill of health.
Cheers,
Michael
For the record: This is before the patch on Itanium SLES 10:
sage -t devel/sage/sage/graphs/graph_isom.pyx **********************************************************************
File "/home/mabshoff/sage-3.0.4.alpha2-SLES10-4.3.1/tmp/graph_isom.py", line 1383:
sage: Y0.is_isomorphic(Y1)
Expected:
True
Got:
False
**********************************************************************
File "/home/mabshoff/sage-3.0.4.alpha2-SLES10-4.3.1/tmp/graph_isom.py", line 1385:
sage: Y0.is_isomorphic(HS)
Expected:
True
Got:
False
**********************************************************************
1 items had failures:
But with the patch applied it still fails:
mabshoff@iras:~/sage-3.0.4.alpha2-SLES10-4.3.1> ./sage -t -long devel/sage-main/sage/graphs/graph_isom.pyx
sage -t -long devel/sage-main/sage/graphs/graph_isom.pyx **********************************************************************
File "/home/mabshoff/sage-3.0.4.alpha2-SLES10-4.3.1/tmp/graph_isom.py", line 1383:
sage: Y0.is_isomorphic(Y1)
Expected:
True
Got:
False
**********************************************************************
File "/home/mabshoff/sage-3.0.4.alpha2-SLES10-4.3.1/tmp/graph_isom.py", line 1385:
sage: Y0.is_isomorphic(HS)
Expected:
True
Got:
False
**********************************************************************
1 items had failures:
2 of 124 in __main__.example_24
***Test Failed*** 2 failures.
For whitespace errors, see the file /home/mabshoff/sage-3.0.4.alpha2-SLES10-4.3.1/tmp/.doctest_graph_isom.py
[89.4 s]
exit code: 1024
So this is likely a gcc bug or something that exposes the issue.
Cheers,
Michael
The patch itself fixes a problem and should be applied obviously. I meant to mention that on the ticket.
Cheers,
Michael
Merged in Sage 3.0.4.rc2
Component: memleak
Issue created by migration from https://trac.sagemath.org/ticket/3607