ParaView issueshttps://gitlab.kitware.com/paraview/paraview/-/issues2020-07-24T10:47:32-04:00https://gitlab.kitware.com/paraview/paraview/-/issues/19614XDMF reader fails to read attribute data when running in parallel2020-07-24T10:47:32-04:00Paul FerrellXDMF reader fails to read attribute data when running in parallelReproducing:
1. Run paraview with more than one pvserver.
2. Open the given xdmf file using the 'XDMF Reader'.
3. Click 'Apply'.
It works fine with a single pvserver instance and using just the client.
[test.hdf5](/uploads/f7dddeae4...Reproducing:
1. Run paraview with more than one pvserver.
2. Open the given xdmf file using the 'XDMF Reader'.
3. Click 'Apply'.
It works fine with a single pvserver instance and using just the client.
[test.hdf5](/uploads/f7dddeae404f72ef32afc6455769ff29/test.hdf5)
[test.xdmf](/uploads/3cfdf4225bd56eed45e8060651cb5b49/test.xdmf)
```
ERROR: In [test.hdf5](/uploads/11f56ec6a8e82d71f7072cc432199581/test.hdf5)
[test.xdmf](/uploads/80a4a7393ecff70a877e0cb716035f95/test.xdmf)5.6.1-osmesa_intel-18.0.5_openmpi-2.1.2/superbuild/paraview/src/VTK/IO/Xdmf2/vtkXdmfHeavyData.cxx, line 1214
vtkXdmfReader (0x36f3640): Failed to read attribute data
```Cory Quammencory.quammen@kitware.comCory Quammencory.quammen@kitware.comhttps://gitlab.kitware.com/paraview/paraview/-/issues/19608XDMF3 RectilinearGrid with HyperSlabbed coordinate arrays are not working.2020-07-24T10:48:13-04:00David E. DeMarleXDMF3 RectilinearGrid with HyperSlabbed coordinate arrays are not working.Files like this:
<Grid GridType="Uniform" Name="0">
<Topology Type="2DRectMesh" NumberOfElements="5 5"/>
<Geometry Type="VXVY">
<DataItem ItemType="HyperSlab" Dimensions="5" Type="HyperSlab" ><DataItem Dimensio...Files like this:
<Grid GridType="Uniform" Name="0">
<Topology Type="2DRectMesh" NumberOfElements="5 5"/>
<Geometry Type="VXVY">
<DataItem ItemType="HyperSlab" Dimensions="5" Type="HyperSlab" ><DataItem Dimensions="3 2" NumberType="Int" Format="XML">0 0 1 1 1 5</DataItem>
<DataItem Format="HDF" Dimensions="4 5" Name="x" NumberType="Float" Precision="8">
test.hdf5:/Locations/x</DataItem>
</DataItem>
<DataItem ItemType="HyperSlab" Dimensions="5" Type="HyperSlab" ><DataItem Dimensions="3 2" NumberType="Int" Format="XML">0 0 1 1 1 5</DataItem>
<DataItem Format="HDF" Dimensions="4 5" Name="y" NumberType="Float" Precision="8">
test.hdf5:/Locations/y</DataItem>
</DataItem>
</Geometry>
</Grid>
Are were found to work in Xdmf2, but not in Xdmf3 such that the X and Y coordinate arrays are nonsensical when read. It should work in either generation of the reader.Cory Quammencory.quammen@kitware.comCory Quammencory.quammen@kitware.comhttps://gitlab.kitware.com/paraview/paraview/-/issues/13978xdmf -binary files crashes with 7 time steps or more.2020-05-04T06:40:22-04:00Kitware Robotxdmf -binary files crashes with 7 time steps or more.**This issue was created automatically from an original [Mantis Issue](http://paraview.org/Bug/view.php?id=13978). Further discussion may take place here.**
---
When I wrap xdmf around numerous time variable binary files, ParaView ...**This issue was created automatically from an original [Mantis Issue](http://paraview.org/Bug/view.php?id=13978). Further discussion may take place here.**
---
When I wrap xdmf around numerous time variable binary files, ParaView crashes when deleting this file.
Linux, local server, master (and 3.98.1). Test dataset is one I voodoo'd up. It basically has 8 4 byte floats that are 0 (i.e., 0x00000000). Dataset is attached, and is called stl-timestep-crash.
* Open crash-opening-seven.xmf.
* Delete crash-opening-seven.xmf.
It will crash.
https://gitlab.kitware.com/paraview/paraview/-/issues/20951XDMF reader produces error "vtkImageSliceRepresentation(0x19ffe1f0) is of typ...2021-09-10T09:12:06-04:00baloeXDMF reader produces error "vtkImageSliceRepresentation(0x19ffe1f0) is of type vtkRectilinearGrid, but a vtkImageData is required"When loading state files where an xmf-file is referenced I keep getting the error message
```
vtkDemandDrivenPipeline:760 ERR| vtkPVDataRepresentationPipeline (0x1a96d3c0): Input for connection index 0 on input port index 0 for algor...When loading state files where an xmf-file is referenced I keep getting the error message
```
vtkDemandDrivenPipeline:760 ERR| vtkPVDataRepresentationPipeline (0x1a96d3c0): Input for connection index 0 on input port index 0 for algorithm vtkImageSliceRepresentation(0x19ffe1f0) is of type vtkRectilinearGrid, but a vtkImageData is required.
```
Even without ever creating anything other than the xdmf.
It is really annoying, but at least everything still seems to functions right.https://gitlab.kitware.com/paraview/paraview/-/issues/20549Unable to visualise XDMF checkpoint in 1D2021-03-11T09:14:19-05:00Rémi Delaporte-MathurinUnable to visualise XDMF checkpoint in 1DWhen using checkpoint XDMF in FEniCS for 1D domains, the results can't be visualised in Paraview (latest version).
The file can be opened with Paraview but no "Plot over line" is available.
![image](/uploads/4eb5e13fd99161b55f365306a2c...When using checkpoint XDMF in FEniCS for 1D domains, the results can't be visualised in Paraview (latest version).
The file can be opened with Paraview but no "Plot over line" is available.
![image](/uploads/4eb5e13fd99161b55f365306a2c29818/image.png)
Example files:
[u_checkpoint.xdmf](/uploads/156e4331d2a9dce2888f4040fc6d00d0/u_checkpoint.xdmf)
[u_checkpoint.h5](/uploads/8650ae4aff8e3bfa574ccbe1c6c1afe5/u_checkpoint.h5)
This issue has been discussed here: https://discourse.paraview.org/t/unable-to-visualise-xdmf-checkpoint-in-1d/2781/12https://gitlab.kitware.com/paraview/paraview/-/issues/20380XDMF: Save -> Load -> Crash2020-12-11T06:41:52-05:00Ken SheldonXDMF: Save -> Load -> CrashObserved in Paraview 5.9.0-RC2 on Windows
1. Create a point set (Sources -> alphabetical -> Point Source, #npoints=100, radius=5)
2. Save data using xdmf or xdmf3 writer
3. Disconnect (delete all data sources)
4. Load the data saved in ...Observed in Paraview 5.9.0-RC2 on Windows
1. Create a point set (Sources -> alphabetical -> Point Source, #npoints=100, radius=5)
2. Save data using xdmf or xdmf3 writer
3. Disconnect (delete all data sources)
4. Load the data saved in step 2.
crash
It makes no difference which writer or which reader you choose.https://gitlab.kitware.com/paraview/paraview/-/issues/20055Problem Importing XDMF file using Open Data With : XDMF Reader2020-07-02T22:22:37-04:00alfredo vanelliProblem Importing XDMF file using Open Data With : XDMF ReaderHi,
I've a problem importing in ParaView5.5.0 and ParaView5.6.0 (that use HDF5 1.8.13) or in ParaView5.8.0 (that use HDF5 1.10.5) the attacched NewLayerModel.h5.xmf XDMF file, loaded using XDMF Reader. File is loaded, but in stdout I've ...Hi,
I've a problem importing in ParaView5.5.0 and ParaView5.6.0 (that use HDF5 1.8.13) or in ParaView5.8.0 (that use HDF5 1.10.5) the attacched NewLayerModel.h5.xmf XDMF file, loaded using XDMF Reader. File is loaded, but in stdout I've got the error messages reported in the attached stdout.txt file. Moreover looking at the Information panel (see attacched screenshotPV5.6.PNG file) I can see that the Cell_K_index and LayerID variables are not loaded correctly. The problem seems due to the use of the CellsOne DataItem defined in the .xmf file and used to create the Cell_K_index and LayerID variables, only when in the HDF5 file the Sebe/Models/Model_0/Input/Events/CellAttributes/CellIindexes dataset cointains a single value (see screenshotHDFVIEW.PNG).
I've verified that the attached deleteme.h5.xmf, where CellIinedexes contains 135 values, is loaded without any problem.
[NewLayerModel.h5](/uploads/385db4d7704fd939ec41537acc7522bd/NewLayerModel.h5)
[NewLayerModel.h5.xmf](/uploads/e810101b14c02991f7fb5b6131458de9/NewLayerModel.h5.xmf)
![screenshotHDFVIEW](/uploads/40ac299ddd23a8b37602ae6e9469186a/screenshotHDFVIEW.PNG)
![screenshotPV5.6](/uploads/8db4c61dc57890403708e5a4bb9a19ec/screenshotPV5.6.PNG)
[stdout.txt](/uploads/c8f77da731f9be3f60117f60745adb9c/stdout.txt)
[deleteme.h5](/uploads/b912150f267031d18c32537c68317632/deleteme.h5)
[deleteme.h5.xmf](/uploads/f994bc9631e58eb7453cbe9791f4800d/deleteme.h5.xmf)https://gitlab.kitware.com/paraview/paraview/-/issues/19904XDMF writer: Version not clear (xdmf vs xdmf3)2020-05-06T11:38:50-04:00Martin Diehlmartin.diehl@kuleuven.beXDMF writer: Version not clear (xdmf vs xdmf3)Steps to reproduce: From the menu, select Sources --> Plane Set X Resolution 10 Set Y Resolution 10 Apply
* Save Data File format Xdmf data file File name SAMPLE.XMF
* Save Data File format Xdmf3 data file File name SAMPLE3.XMF
Both fi...Steps to reproduce: From the menu, select Sources --> Plane Set X Resolution 10 Set Y Resolution 10 Apply
* Save Data File format Xdmf data file File name SAMPLE.XMF
* Save Data File format Xdmf3 data file File name SAMPLE3.XMF
Both files are identical and contain `<Xdmf xmlns:xi="http://www.w3.org/2001/XInclude" Version="3.0">`. Also, when for both writers the dialogue popping up (Choose Arrays To Write/Write All Time Steps) says "Configure Write (Xdmf3Writer)"
Tested with Paraview 5.8 on Windows 10 (64 bit)https://gitlab.kitware.com/paraview/paraview/-/issues/19903XDMF Writer crashes w/o write permission2020-05-06T11:36:16-04:00Martin Diehlmartin.diehl@kuleuven.beXDMF Writer crashes w/o write permissionSteps to reproduce: From the menu, select Sources --> Plane Set X Resolution 10 Set Y Resolution 10 Apply File --> Save Data File format Xdmf data file File name C:\\SAMPLE.XMF
As a normal user, I don't have write permission. Instead of...Steps to reproduce: From the menu, select Sources --> Plane Set X Resolution 10 Set Y Resolution 10 Apply File --> Save Data File format Xdmf data file File name C:\\SAMPLE.XMF
As a normal user, I don't have write permission. Instead of getting an error message, Paraview silently crashes. This happens with both writers (xdmf3 and xdmf)
Paraview 5.8.0 64 Bit for Windows 10 (64-bit)https://gitlab.kitware.com/paraview/paraview/-/issues/19607Only last XDMF FiniteElementFunction attribute is visible in ParaView2020-05-07T12:31:49-04:00Darren ParfittOnly last XDMF FiniteElementFunction attribute is visible in ParaViewWhen I have two XDMF attributes using the ItemType="FiniteElementFunction" inside a single grid object ParaView only lists the last one in its available variables to plot. I have uploaded the example file I used. This may be an issue wi...When I have two XDMF attributes using the ItemType="FiniteElementFunction" inside a single grid object ParaView only lists the last one in its available variables to plot. I have uploaded the example file I used. This may be an issue with ParaView or lower down with the actual VTK/XDMF3Reader.
The example include two cell centred results to show these work. I would have included nodal as well but if you do that you hit another bug (discussed [here](https://discourse.paraview.org/t/xdmf3-reader-finiteelementfunction/3079) ). This may be related because in my case it complains about the nodal result having only 6 values (which it should) when PV is expecting 8 (which is actually the size of the FiniteElement attributes)
I have tried PV 5.5.2, 5.6.1 and 5.7 and they all show the same issue.
[test_finiteelement.xmf](/uploads/654153ebd082900fe29ce82e1f846240/test_finiteelement.xmf)https://gitlab.kitware.com/paraview/paraview/-/issues/19514Support of Polyhedral Cells with XDMF 32020-11-19T09:52:00-05:00Roman FuchsSupport of Polyhedral Cells with XDMF 3There are 3 readers for XDMF files in Paraview: XDMF Reader, Xdmf3ReaderS, Xdmf3ReaderT. But there is no information how they differ among each other.
I noticed that:
- XDMF Reader: cannot read polyhedral cells, can read grid-of-grids
...There are 3 readers for XDMF files in Paraview: XDMF Reader, Xdmf3ReaderS, Xdmf3ReaderT. But there is no information how they differ among each other.
I noticed that:
- XDMF Reader: cannot read polyhedral cells, can read grid-of-grids
- Xdmf3ReaderS: can read polyhedral cells, cannot read grid-of-grids
Reading a polyhedral grid with Xdmf3ReaderS, I also noted that the time tag is not properly imported.
The XDMF website says that the tag "NumberOfElements" is only for clarity; however, Paraview needs that value at least for a few cell types (polyvertex, polyline, polygon) otherwise it crashes.
Clarify/extend your documentation and ensure that all data interfaces run as expected.https://gitlab.kitware.com/paraview/paraview/-/issues/19408Selecting subset of variables with XDMFReader does not work with series of files2021-03-24T06:06:38-04:00Alexander ChadwickSelecting subset of variables with XDMFReader does not work with series of filesIf I try to read in the entire series of XDMF files with the XDMF Reader (i.e., XDMF2 instead of XDMF3), if I select only some of the variables in the dataset, upon clicking apply ParaView ignores the unchecked variables and still appear...If I try to read in the entire series of XDMF files with the XDMF Reader (i.e., XDMF2 instead of XDMF3), if I select only some of the variables in the dataset, upon clicking apply ParaView ignores the unchecked variables and still appears to read in all of the variables. The issue is minor (having to switch variables I'm coloring by, having to hide extra colorbars when scripting) to severe (massive memory usage with larger datasets). It does not occur if I read in only a single XDMF file at a time or if I use XDFM3T (latter does not have as good parallel performance).
I encounter this issue in the official ParView 5.7.0 binaries for Mac installed via the .pkg distribution on a 2013 Mac Pro running Mojave 10.14.6. The test dataset contains four variables that were written to individual HDF5 archives for each process. The whole dataset is then stitched using the HDF5 virtual dataset feature, with that HDF5 archive being referenced in the XDMF header.
To reproduce:
1. Open the entire iter_*.xdmf set of files (two total)
2. Select XDMF Reader
3. In Cell Arrays, uncheck "ArrayIDontWant1" and "VectorArrayIDontWant1" and click apply
4. Click on coloring menu to change from "Solid Color." Unchecked variables are somehow still visible
Attached: [xdmf_test.zip](/uploads/0cef498a44ce2ec538e6dbe1c281e339/xdmf_test.zip)https://gitlab.kitware.com/paraview/paraview/-/issues/18889Segfault with pvbatch (ParaView 5.6.0) with use of GetDataInformation() with ...2020-05-07T12:32:40-04:00Emanuele LeggioSegfault with pvbatch (ParaView 5.6.0) with use of GetDataInformation() with the Xdmf3ReaderWith the following simple script pvbatch crashes with segfault before `print 'point 2'`:
```python
from paraview.simple import *
obj1 = Xdmf3ReaderT(FileName=['cone.xmf'])
obj1.UpdatePipeline()
a = obj1.GetDataInformation()
obj2 = X...With the following simple script pvbatch crashes with segfault before `print 'point 2'`:
```python
from paraview.simple import *
obj1 = Xdmf3ReaderT(FileName=['cone.xmf'])
obj1.UpdatePipeline()
a = obj1.GetDataInformation()
obj2 = Xdmf3ReaderT(FileName=['cone.xmf'])
obj2.UpdatePipeline()
print 'point 1'
b = obj2.GetDataInformation()
print 'point 2'
```
cone.xmf is a simple source cone saved as Xdmf3 data file.
Same behavior with different files instead of cone.xmf or with different reading files for obj1 and for obj2.https://gitlab.kitware.com/paraview/paraview/-/issues/18763Segfault when contouring h5+xdmf with N>1 pvservers2021-03-23T19:13:23-04:00Erik KeeverSegfault when contouring h5+xdmf with N>1 pvserversWhen visualizing 3D a structured grid time series in PV, while connected to more than 1 pvserver, one pvserver rapidly (but at random) crashes with a sigsegv. I have uploaded the complete test dataset I am playing with here: http://uoreg...When visualizing 3D a structured grid time series in PV, while connected to more than 1 pvserver, one pvserver rapidly (but at random) crashes with a sigsegv. I have uploaded the complete test dataset I am playing with here: http://uoregon.edu/~ekeever1/testset.tar.gz (apologies for the 70M test set size).
When running in serial (just paraview) or connected to one pvserver, I can run the visualization back and forth for a long time without issue.
Steps to reproduce for me:
- Connect to N > 1 pvservers
- Open test_metadata.xdmf
- Apply
- Create contour of mass, uncheck 'compute ..'
- Delete default contour value, create range with defaults, set log scale, ok, apply
- Play animation repeatedly: Segfault happens eventually
Output to console with gdb not attached:
[segv_output](/uploads/3c3149093d2dfeb306903fa3698bb99e/segv_output)
With GDBs attached to all 4 pvservers, I catch the crash. The backtrace from the offending process within gdbtui:
[gdboutput](/uploads/3cf1ce00c93536df3e3da91524bde88c/gdboutput)
The core of the offending process (warning, 340MB): http://uoregon.edu/~ekeever1/core.21607
#0 0x00007f2c5c98cf8f in vtkAOSDataArrayTemplate<float>::GetTuple (this=0x557f58461450, tupleIdx=1059818420, tuple=0x7ffdfb6f5540)
at /home/erik-k/manual_install/ParaView-v5.6.0/VTK/Common/Core/vtkAOSDataArrayTemplate.txx:252
The evident enough cause:
```
(gdb) print i
$6 = 0
(gdb) info locals
i = 0
data = 0x55824dce3b90
(gdb) print data[i]
Cannot access memory at address 0x55824dce3b90
(gdb) print this->Buffer->GetBuffer()
$7 = (vtkBuffer<float>::ScalarType *) 0x557f57c3af20
(gdb) print tupleIdx
$8 = 1059818420
```
Well, that'll do it: The test data set is 64x64x24, only ~100K cells...
Up frame:
void GetPoint(vtkIdType id, double x[3])
VTK_EXPECTS(0 <= id && id < GetNumberOfPoints())
VTK_SIZEHINT(3)
{ this->Data->GetTuple(id,x); }
```
#1 0x00007f2c61886f5d in vtkPoints::GetPoint (this=0x557f587e0a10, id=1059818420, x=0x7ffdfb6f5540) at /home/erik-k/manual_install/ParaView-v5.6.0/VTK/Common/Core/vtkPoints.h:146
(gdb) print id
$10 = 1059818420
(gdb) print *x@3
$13 = {0.4330126941204071, 0.25, 0}
```
This corresponds to a point on the innermost rim of the dataset for what it's worth.
```
(gdb) up
#2 0x00007f2c5dc438f6 in vtkPolyData::ComputeBounds (this=0x557f58ce5490) at /home/erik-k/manual_install/ParaView-v5.6.0/VTK/Common/DataModel/vtkPolyData.cxx:603
(gdb) info locals
i = 1
pts = 0x557f59709b18
cella = {0x557f56877e90, 0x557f56877e90, 0x557f59082360, 0x557f56877e90}
t = 2
npts = 3
x = {0.4330126941204071, 0.25, 0}
doneOne = 1
(gdb) print *pts@4
$21 = {4923, 1059818420, 4922, 0}
```
--------
Hope this report is useful.https://gitlab.kitware.com/paraview/paraview/-/issues/17945Segfault when reading XDMF mesh2020-05-05T23:15:36-04:00Nico SchlömerSegfault when reading XDMF meshI'm not sure if this is a ParaView, a VTK, or an XDMF issue. Anyhow, for a segfault it's probably always worth filing a bug.
Steps to reproduce:
* Open
[mesh.xdmf](/uploads/03294831220e1cbe2a15123c963a5529/mesh.xdmf)
[mesh.h5](...I'm not sure if this is a ParaView, a VTK, or an XDMF issue. Anyhow, for a segfault it's probably always worth filing a bug.
Steps to reproduce:
* Open
[mesh.xdmf](/uploads/03294831220e1cbe2a15123c963a5529/mesh.xdmf)
[mesh.h5](/uploads/bf0bc7534c6ceb97c7c519d7f0a8a188/mesh.h5)
with ParaView.
* Choose `XDMF3 Reader` or `XDFM3 Reader (Top Level Partition)`.
* Click `Apply`.
* Behold `Segmentation fault (core dumped)`.
Interestingly, when the same data is given as XML-XDMF [mesh-xml.xdmf](/uploads/3853fb99837d8fc6f5b6b454449fffbb/mesh-xml.xdmf), no such error occurs.https://gitlab.kitware.com/paraview/paraview/-/issues/17925Parse error not caught when reading xml file2020-05-05T02:07:42-04:00Denis-Gabriel CapraceParse error not caught when reading xml fileWhen feeding a XDMF3 reader with a corrupted xml file, in my case associated to a hdf5 file, Paraview simply crashes while trying to read it, perhaps because the parse error thrown by the reader is not caught.When feeding a XDMF3 reader with a corrupted xml file, in my case associated to a hdf5 file, Paraview simply crashes while trying to read it, perhaps because the parse error thrown by the reader is not caught.https://gitlab.kitware.com/paraview/paraview/-/issues/17871Paraview crashes with XDMF3 Reader2020-05-07T04:18:57-04:00Loïc BertheParaview crashes with XDMF3 ReaderHi,
When I try to open XDMF files, it crashes when I use the XDMF3 Reader.
It seems to load fine with the classical "XDMF Reader".
I've had this issue with the following version of ParaView :
- Paraview 5.4.1 under Windows 7
- Paraview...Hi,
When I try to open XDMF files, it crashes when I use the XDMF3 Reader.
It seems to load fine with the classical "XDMF Reader".
I've had this issue with the following version of ParaView :
- Paraview 5.4.1 under Windows 7
- Paraview 5.0.1 under Linux 64
Could you clarify which limitations come with the new XDMF3 reader ?
You can find below a test case reproducing that issue.
Best regards,
Loïc
[data.xdmf](/uploads/e76430df35856c36e55d4232f5f10103/data.xdmf)
[data.h5](/uploads/4b5978a598d5c1e80ae0f1ec06fbae7b/data.h5)https://gitlab.kitware.com/paraview/paraview/-/issues/17720Xdmf2: 3-component arrays saved to XDMF and re-read do not have 3 components2021-03-23T13:56:19-04:00Cory Quammencory.quammen@kitware.comXdmf2: 3-component arrays saved to XDMF and re-read do not have 3 componentsSteps to reproduce:
* Launch ParaView/builtin server
* Sources -> Sphere
* File -> Save Data. Save as an Xdmf Data File (*.xmf)
* Reset session.
* Load the previously saved Xdmf file. The Normals array from the Sphere source will be pre...Steps to reproduce:
* Launch ParaView/builtin server
* Sources -> Sphere
* File -> Save Data. Save as an Xdmf Data File (*.xmf)
* Reset session.
* Load the previously saved Xdmf file. The Normals array from the Sphere source will be present, but it will have a single component.
The attached files are from the Xdmf file export in this series of steps.
[VectorTest.xmf](/uploads/7ae7e1ce33d2eb5d509fd6691a231e6b/VectorTest.xmf)
[VectorTest.h5](/uploads/229935fa7d6a01315e47bd0f1e5fa451/VectorTest.h5)
Note that the `Attribute` element has an XML attribute named `Type` which is set to `"None"`. Changing this to an XML attribute named `AttributeType` and setting its value to `"Vector"` causes the "Normals" array to be read in correctly as a vector.https://gitlab.kitware.com/paraview/paraview/-/issues/17525XDMF: Various issues with hyperslab and XML dataset (xdmf2 and xdmf3 readers)2020-05-05T03:14:41-04:00Armin WehrfritzXDMF: Various issues with hyperslab and XML dataset (xdmf2 and xdmf3 readers)When trying to debug or workaround the issues raise in #17516 #17295 #17166 #17517 I stumbled across various other issues. I attached a few data-files that I believe should all be valid XDMF files (all based on a XDMF dataset generated w...When trying to debug or workaround the issues raise in #17516 #17295 #17166 #17517 I stumbled across various other issues. I attached a few data-files that I believe should all be valid XDMF files (all based on a XDMF dataset generated with the wavelet source in ParaView 5.4).
Essentially only the original dataset works with both readers (xdmf2 and xdmf3). Adding additional functionality (e.g. hyperslab dataset) or changing the dataset to TXT/XML using `xi:include` does not work or leads to crashes (segfaults) with one or both readers.
XML `xi:include` issues:
1. [wavelet_09x09x09.xmf](/uploads/725c38cc6475de100b9387447e4fb636/wavelet_09x09x09.xmf) combined with [wavelet_09x09x09.h5](/uploads/849f4763fb52949ab3e806cb5b0f60a8/wavelet_09x09x09.h5) works with both readers as expected, as it is the original dataset.
2. [wavelet_09x09x09_xml.xmf](/uploads/ddeb5c1a927263148af23aaaf1de2898/wavelet_09x09x09_xml.xmf) combined with [wavelet_09x09x09.txt](/uploads/3b8e666bb8ecdfe6c1c2ffdcd10e5c9e/wavelet_09x09x09.txt) using `xi:include` to insert the dataset in ascii format from a text file works with the xdmf2 reader correctly but crashes with the xdmf3 reader: ```parser error : Start tag expected, '<' not found```.
It seems xdmf3 does not consider the ```parse='text'``` attribute.
Hyperslab issues:
1. [wavelet_05x05x09_hyperslab.xmf](/uploads/c241074fca9b2f1fda82224a0775175f/wavelet_05x05x09_hyperslab.xmf) works correctly with the xdmf2 reader, but the xdmf3 does not read the dataset issuing the following warning:
```
Skipping unrecognized array type [None]
```
This appears to be the same bug as in issue #17516 since the error message and behaviour is the same.
2. [wavelet_05x05x09_hyperslab_xml.xmf](/uploads/3e4a2fe3a6495d07d7a0118b96222f3e/wavelet_05x05x09_hyperslab_xml.xmf) does not work with either one of the readers though I believe for different reasons. This xdmf file is a combination of a hyperslab and a dataset in XML format. Each of these features works with the xdmf2 reader (though not with the xdmf3) and the combination crashes with the both.
* xdmf2: `realloc(): invalid next size: 0x0000000006a9b5b0 ***` ... `Segmentation fault`
I also tried to copy&past the content of the text file directly into the xmf file, but the error was the same. Thus I conclude it's not an issue with the `xi:include` as this works with the xdmf2 reader otherwise.
* xdmf3: ```parser error : Start tag expected, '<' not found```.
Obviously the xdmf3 doesn't even read the dataset and I can only assume that the hyperslab wouldn't work either.
The .h5 and .txt files are used in various .xmf files, thus all files should be downloaded and put to the same directory. Here a list of all dataset:
* [wavelet_09x09x09.xmf](/uploads/725c38cc6475de100b9387447e4fb636/wavelet_09x09x09.xmf)
* [wavelet_09x09x09.h5](/uploads/849f4763fb52949ab3e806cb5b0f60a8/wavelet_09x09x09.h5)
* [wavelet_09x09x09_xml.xmf](/uploads/ddeb5c1a927263148af23aaaf1de2898/wavelet_09x09x09_xml.xmf)
* [wavelet_09x09x09.txt](/uploads/3b8e666bb8ecdfe6c1c2ffdcd10e5c9e/wavelet_09x09x09.txt)
* [wavelet_05x05x09_hyperslab.xmf](/uploads/c241074fca9b2f1fda82224a0775175f/wavelet_05x05x09_hyperslab.xmf)
* [wavelet_05x05x09_hyperslab_xml.xmf](/uploads/3e4a2fe3a6495d07d7a0118b96222f3e/wavelet_05x05x09_hyperslab_xml.xmf)
All testing was done with ParaView 5.4 (binary installer: ParaView-5.4.0-Qt5-OpenGL2-MPI-Linux-64bit.tar.gz)https://gitlab.kitware.com/paraview/paraview/-/issues/17517XDMF2 fails to concatenate more than 10 arrays in a Function DataItem2020-05-05T03:15:54-04:00Armin WehrfritzXDMF2 fails to concatenate more than 10 arrays in a Function DataItemXDMF2 fails to concatenate more than 10 arrays in a Function DataItem. Attached are two files, where only `wavelet_9x2+9.xmf` is correctly read. The second file (`wavelet_12x2+3.xmf`) with more than 10 concatenated dataset fails with the...XDMF2 fails to concatenate more than 10 arrays in a Function DataItem. Attached are two files, where only `wavelet_9x2+9.xmf` is correctly read. The second file (`wavelet_12x2+3.xmf`) with more than 10 concatenated dataset fails with the following error:
```
XdmfExpr : syntax error
Segmentation fault (core dumped)
```
None of the files can be read with the xdmf3 reader, see #17516. (The xdmf3 reader also cannot handle XPath references which prevents me from using it altogether for my actual dataset, see #17166).
Ping: @demarle @AndrewBurns
[wavelet_9x2+9.xmf](/uploads/c231cdeba75768fc09ea9b336b30d082/wavelet_9x2+9.xmf)
[wavelet_12x2+3.xmf](/uploads/f81e2ee3df94783d519e65e5752a1eee/wavelet_12x2+3.xmf)