[Paraview] Named blocks in pvd Collection with time series

[Giuseppe Bilotta]

Hello all,

I have the following situation: I have multiple dataset created by the
same simulation, evolving in time. I'm currently presented wit two
options to preserve time information:

Option 1: I create a pvd in the form:

<?xml version='1.0'?>
<VTKFile type='Collection' version='0.1'>
<Collection>
<DataSet timestep='0' part='1' name='Planes' file='PLANES.vtu'/>
<DataSet timestep='0' part='2' name='WaveGages' file='WaveGage_00000.vtu'/>
<DataSet timestep='0' part='3' name='Particles' file='PART_00000.vtu'/>
<DataSet timestep='0.00049' part='1' name='Planes' file='PLANES.vtu'/>
<DataSet timestep='0.00049' part='2' name='WaveGages'
file='WaveGage_00001.vtu'/>
<DataSet timestep='0.00049' part='3' name='Particles' file='PART_00001.vtu'/>
<!-- etc for the other timesteps -->
</Collection>
</VTKFile>

which is efficient (single index file) but annoying UI-wise (since the
names of the data sets are not used in the muti-block inspector or in
the extract block filter)

Option 2: I have the pvd point, for each timestep, to a single .vtm
file which in turns points to the multiple vtu files for that
timestep: so the pvd is like

<?xml version='1.0'?>
<VTKFile type='Collection' version='0.1'>
<Collection>
<DataSet timestep='0' group='' part='0' file='data_00000.vtm'/>
<DataSet timestep='0.00049' group='' part='0' file='data_00001.vtm'/>
</Collection>
</VTKFile>

and each vtm is like:

<?xml version='1.0'?>
<VTKFile type='vtkMultiBlockDataSet' version='1.0'>
<vtkMultiBlockDataSet>
<DataSet index='0' name='Planes' file='PLANES.vtu'/>
<DataSet index='1' name='WaveGages' file='WaveGage_00000.vtu'/>
<DataSet index='2' name='Particles' file='PART_00000.vtu'/>
</vtkMultiBlockDataSet>
</VTKFile>

This has the advantage that the name of each dataset appears in the
block extraction tool, but the disadvantage that I have to create one
additional file per timestep (which may be a LOT of extra files)

I've also tried Option 3: a pvd pointing to one pvd per block, and
each pvd having the time series for that block, but this simply
segfaults paraview.

So my question is: is there a way to combine the compactness offered
Option 1 with the possibility to see the block names in the UI offered
by Option 2? I've also tried to play around with the group option in
the pvd dataset entries, to no effect (and I'm still not clear on the
difference between group and part, so any clarification on this is
also welcome).

Thanks in advance and best regards,


-- 
Giuseppe "Oblomov" Bilotta