CellGeometryDataDump error

[rwendner]

Recently, the CellGeometryDataDump generates inconsistent files.

In the attached sample for 'Cell 72' the given number of external edges is 'nxe: 201'.
Yet, 202 lines follow before 'Cell 73'.

This causes every reader to crash and is a disaster for postprocessing.

[zhouxinwei]

I couldn't reproduce the issue. Can you please attach the LDPM mesh file instead of the Gen file? It seems I am getting a different mesh somehow.

[zhouxinwei]

The bug (fixed in version 2016.8.10) was activated when you make the element selections as below, maybe you didn't do this before so no issue occured.

When dumping the cell data, all elements should be selected, see below:

Code: Select all

TetSolidList PRTC {
  EditNodeList {
    Select all
    Select cx < 0.03 m
    Reselect cx > 0.0 m
    ...
    Make NodeList SelNode
  }
  Select 1n
  Make TetSolidList SelPart
  Select All // add this line to restore default status, which is selected. This will fix the issue.
}


Manual is updated to remind users.

[rwendner]

Thank you for this explanation.

However, I seriously disagree.
The results of a paraview plot or datadump should NOT be affected by a random selection somewhere in the file.
The only selection that should have an affect (if at all for a dumpfile) is the active selection at the time when CellGeometryDataDump or FacetGeometryDataDump is called.
Otherwise, the behavior of MARS becomes unpredictable.

Independent of this more general discussion on the affect of selections I still have the problem.
In the attached example, I have the suggested "select all" command. However, I still get the erroneous dumpfile as described above.

[zhouxinwei]

You need to use version 2016.8.10. This version will check the issue reported, if it occurs, MARS will error off.

[rwendner]

I just tried with version 2016.8.10 and get a error message "something is wrong in LDPMCell::dataDump()

[zhouxinwei]

If I was not clear, you need to use the latest version and add "Select All".

[rwendner]

I tried with and without 'Select all'

without 'select all' I get the error message
with 'select all' I don't get it, the dump file is generated but I still can not read it
(the reader crashed in the same line as before, trying to read Cell 73 but Cell 72 has one line too much in the nxe section

[zhouxinwei]

Since version 2016.8.12, all lists are restored to default status, which is selected, after the input file is read, the users no longer need to add "Select All" to restore selection status.

Just to clarify, in version 2016.8.10, the "something is wrong in LDPMCell::dataDump()" message was checking exactly the issue you were reporting. If you don't see this message after adding "Select All", you should have no problem. Were you, by any change, checking the old file?

Also, as I mentioned before, I couldn't reproduce your mesh file, maybe you were using a different version of tetgen.

[rwendner]

I tried many times (evertime deleting the previous dump files) - always the same result.
Now I restarted everything (windows approach) and can't reproduce the problem anymore.

Thank you for fixing it.