Alembic Stitch Example Segfaults

[GoogleCodeExporter]

What steps will reproduce the problem?

1. Create a plane subdivided 53x53 or higher (>2730 polys).  I have reproduced 
it leaving the plane completely static as well as animating it's position and 
deforming it.
2. use AbcExport to export the plane into batches over time (e.g. total export 
is 1-10, into 4 files 1-3, 4-6, 7-9, 10)
3. Run the abcstitcher executable and it should seg fault

What is the expected output? What do you see instead?

It should stitch the abc files into one combined file.  Instead it crashes.  
Attached are the batch abc files as well as the failed stitch.  In addition, in 
case it is helpful, the backtrace is attached.

What version of the product are you using? On what operating system?

Alembic 1.0.1 on Linux CentOS 5.3.

Please provide any additional information below.

Lucas was able to reproduce the problem as well.

Original issue reported on code.google.com by shappysf...@gmail.com on 6 Oct 2011 at 6:54

Attachments:

• plane_test.tar.gz
• backtrace_abc_stitcher.log

[GoogleCodeExporter]

The problem is with how the normals are being set in the stitcher, not with the 
number of polygons.

The fix for this is here:
http://code.google.com/r/millerlucas-dev/source/detail?r=04be3465907a98cead8e040
c9ec192e35abd4d04

Original comment by miller.lucas on 7 Oct 2011 at 6:18

• Changed title: Alembic Stitch Example Segfaults
• Changed state: PleaseVerify
• Added labels: Milestone-1.0.2, Priority-P1

[GoogleCodeExporter]

Thanks Lucas!  This fix works.  I tested it for the simple subdivided plane and 
also a much more complex production scene.

Original comment by shappysf...@gmail.com on 7 Oct 2011 at 6:39

[GoogleCodeExporter]

Original comment by miller.lucas on 15 Oct 2011 at 12:30

• Changed state: Verified