MantisBT - ParaView |
| View Issue Details |
|
| ID | Project | Category | View Status | Date Submitted | Last Update |
| 0009729 | ParaView | Bug | public | 2009-10-17 05:13 | 2011-05-02 20:08 |
|
| Reporter | Sven Buijssen | |
| Assigned To | Nathan Fabian | |
| Priority | normal | Severity | minor | Reproducibility | always |
| Status | closed | Resolution | fixed | |
| Platform | | OS | | OS Version | |
| Product Version | 3.6 | |
| Target Version | | Fixed in Version | | |
| Project | |
| Topic Name | |
| Type | |
|
| Summary | 0009729: vtkExodusIIWriter fails to export two flat, convex, simple polygons (n=5 & n=6) correctly => invalid file |
| Description | Load the attached dataset 2-polygons-triggering-exodusIIwriter-error.vtp. Save it in Exodus II File format. The writer fail raise the following error message:
ParaView3/VTK/Parallel/vtkExodusIIWriter.cxx, line 1721
vtkExodusIIWriter (...): The nodes per element doesn't match (6 and 5), data will not be written correctly
This dataset contains merely two polygons. Load the polygons separately and the writer will do its job smoothly.
Group both polygonal datasets, apply Merge Blocks, apply Extract Surface. When saving in Exodus II File format the error message will pop up again.
The resulting file is really invalid, vtkExodusIIReader fails.
Tested with PV 3.6.1 and CVS HEAD. |
| Steps To Reproduce | |
| Additional Information | |
| Tags | No tags attached. |
| Relationships | |
| Attached Files |  2-polygons-triggering-exodusIIwriter-error.vtp (2,185) 2009-10-17 05:13
https://www.vtk.org/Bug/file/7566/2-polygons-triggering-exodusIIwriter-error.vtp
1-polygon-n=5.vtp (1,856) 2009-10-17 05:14
https://www.vtk.org/Bug/file/7567/1-polygon-n%3D5.vtp
1-polygon-n=6.vtp (1,900) 2009-10-17 05:14
https://www.vtk.org/Bug/file/7568/1-polygon-n%3D6.vtp |
|
| Issue History |
| Date Modified | Username | Field | Change |
| 2009-10-17 05:13 | Sven Buijssen | New Issue | |
| 2009-10-17 05:13 | Sven Buijssen | File Added: 2-polygons-triggering-exodusIIwriter-error.vtp | |
| 2009-10-17 05:14 | Sven Buijssen | File Added: 1-polygon-n=5.vtp | |
| 2009-10-17 05:14 | Sven Buijssen | File Added: 1-polygon-n=6.vtp | |
| 2010-05-27 18:00 | Sven Buijssen | Note Added: 0020848 | |
| 2011-04-14 11:17 | Sven Buijssen | Note Added: 0026176 | |
| 2011-04-14 11:32 | Sebastien Jourdain | Assigned To | => Nathan Fabian |
| 2011-04-14 11:32 | Sebastien Jourdain | Status | backlog => tabled |
| 2011-04-26 13:34 | Nathan Fabian | Note Added: 0026258 | |
| 2011-04-26 13:58 | Sven Buijssen | Note Added: 0026260 | |
| 2011-04-26 14:21 | Nathan Fabian | Note Added: 0026261 | |
| 2011-04-26 19:29 | Nathan Fabian | Note Added: 0026263 | |
| 2011-04-26 19:29 | Nathan Fabian | Status | tabled => @80@ |
| 2011-04-26 19:29 | Nathan Fabian | Resolution | open => fixed |
| 2011-05-02 20:08 | Alan Scott | Note Added: 0026399 | |
| 2011-05-02 20:08 | Alan Scott | Status | @80@ => closed |
|
Notes |
|
|
(0020848)
|
|
Sven Buijssen
|
|
2010-05-27 18:00
|
|
Problem persists in PV 3.8.0
Additionally: try loading 2-polygons-triggering-exodusIIwriter-error.vtp and save it in Exodus II File format, choose File - Disconnect. The released binary ParaView 3.8.0 Windows 32 bit will crash (using Windows XP SP3), Linux x86 32 bit won't, neither does a Linux x86 32bit debug build. |
|
|
|
(0026176)
|
|
Sven Buijssen
|
|
2011-04-14 11:17
|
|
Problem is partially resolved by
commit 8f6755dff35cd88dba1f0b23246712d5f0079ef1
Author: Nathan Fabian <ndfabia@sandia.gov>
Date: Tue Dec 14 13:12:15 2010 -0700
ENH: Polygon write works.
as the error message is not raised any more. But ParaView 3.11.1 still segfaults when trying to read the Exodus II file vtkExodusIIWriter.cxx creates from 2-polygons-triggering-exodusIIwriter-error.vtp |
|
|
|
(0026258)
|
|
Nathan Fabian
|
|
2011-04-26 13:34
|
|
|
I have tried both of the options above, but I am able to read the exodus II file without a crash and the polygons look correct. Is this only occurring on Windows and still in the latest version? |
|
|
|
(0026260)
|
|
Sven Buijssen
|
|
2011-04-26 13:58
|
|
The issue described in note 20848 is resolved using the released binary of ParaView 3.10.1 Windows 32 bit. But the segfault described in note 26176 still occurs on Ubuntu 10.04.2 LTS using a ParaView nightly debug build (commit 037bce21974f7f27e7adc1f4cb0cab2aa6f54a37), built-in server, --disable-registry command line option, following these steps:
* File / Open / 2-polygons-triggering-exodusIIwriter-error.vtp. Apply
* File / Save Data / Exodus II File / 2-polygons-triggering-exodusIIwriter-error.ex2. Ok. Leave "Write All Time Steps" unchecked. Ok.
* File / Disconnect
* File / Open / 2-polygons-triggering-exodusIIwriter-error.ex2. Apply.
=> segfault:
Program received signal SIGSEGV, Segmentation fault.
0x00007ffff6dcf7be in vtkDataArrayTemplate<int>::GetValue (this=0x17cd190, id=0) at /ParaView3.git/VTK/Common/vtkDataArrayTemplate.h:126
126 T GetValue(vtkIdType id) { return this->Array[id]; }
The official binary releases of PV 3.10.1 Win 32bit and Linux 64bit segfault trying to read this file, too.
I could upload 2-polygons-triggering-exodusIIwriter-error.ex2 if that would help. |
|
|
|
(0026261)
|
|
Nathan Fabian
|
|
2011-04-26 14:21
|
|
|
Okay I can repeat the problem with 3.10.1, but it doesn't appear in the repository version 3.11.0 It appears to be fixed. Can you try building from the repository source to test? |
|
|
|
(0026263)
|
|
Nathan Fabian
|
|
2011-04-26 19:29
|
|
I had an old checkout. When the exodus reader cache was removed on the 6th this problem returned.
The caller to GetCacheOrRead needs to ensure a Register/UnRegister if there are going to be other calls to GetCacheOrRead as it will destroy the previously returned array. |
|
|
|
(0026399)
|
|
Alan Scott
|
|
2011-05-02 20:08
|
|
Looks fixed to me. The reporter may want to retest, since we have thought it was fixed before.
Tested local server, Windows, trunk. |
|