Hi Stephen,

The error in the streamlines you show there shouldn't have anything to do with contours.  I'd suggest marking it as good in the bisect process and continuing.  That import was fixed in the next commit, I think.


On Tue, Mar 1, 2011 at 12:19 PM, Stephen Skory <stephenskory@yahoo.com> wrote:
Hi All,

using a script like this <http://paste.enzotools.org/show/1519/> I am getting this error below with 3782:f9c66b2cacfc.

 File "/duanestorage/home/faculty/stsk4691/plot_clumps.py", line 26, in <module>
   find_clumps(master_clump, c_min, c_max, step)
 File "/duanestorage/home/faculty/stsk4691/yt/src/yt-hg/yt/analysis_modules/level_sets/clump_handling.py", line 170, in find_clumps
 File "/duanestorage/home/faculty/stsk4691/yt/src/yt-hg/yt/analysis_modules/level_sets/clump_handling.py", line 117, in find_children
 File "/duanestorage/home/faculty/stsk4691/yt/src/yt-hg/yt/analysis_modules/level_sets/contour_finder.py", line 303, in identify_contours
   boundary_tree = amr_utils.construct_boundary_relationships(fd)
 File "ContourFinding.pyx", line 92, in yt.utilities.amr_utils.construct_boundary_relationships (yt/utilities/amr_utils.c:26719)
IndexError: Out of bounds on buffer access (axis 0)

However, I do not get the bug with 3746:1b81f9d8847c, so I ran "hg bisect" in that range. However, part of the way through the bisect process a new bug was introduced:

 File "/duanestorage/home/faculty/stsk4691/yt/lib/python2.6/site-packages/yt-2.1dev-py2.6-linux-x86_64.egg/yt/visualization/api.py", line 59, in <module>
   from streamlines import \
ImportError: cannot import name StreamLines

which polluted the bisect process, I think. The revision that bisect identified as being bad is this one:

The first bad revision is:
changeset:   3766:20a49bc35fbd
branch:      yt
parent:      3763:badcec4be5c9
user:        Sam Skillman <sam.skillman@gmail.com>
date:        Thu Feb 24 19:18:17 2011 -0700
summary:     Adding an initial go at streamlines.  It uses the amr kD-tree to

But like I said, this might not be the culprit. Can anyone think of what they might have done to clump finding (contour_finder.py) in that revision range that could make this first bug above? Thanks!

Stephen Skory
510.621.3687 (google voice)
Yt-dev mailing list