[vtkusers] Possible Bug in vtkXMLReader when reading timeseries

Mon Sep 26 10:39:53 EDT 2016

Simon,

Might you be able to try TJ's merge request and see if it restores the
functionality you expected?

Thanks,
Cory

On Mon, Sep 26, 2016 at 9:34 AM, T.J. Corona <tj.corona at kitware.com> wrote:

> Hi Simon,
>
> I have created a MR (https://gitlab.kitware.com/
> vtk/vtk/merge_requests/2007) that reverts the addition of an
> exit-with-error when data arrays have the same name.
>
> Cory,
>
> I believe this MR also accounts for the segfault you originally reported
> when reading in data arrays that have duplicate names.
>
> Sincerely,
> T.J.
>
> Thomas J. Corona, Ph.D.
> Kitware, Inc.
> R&D Engineer
> 21 Corporate Drive
> Clifton Park, NY 12065-8662
> Phone: 518-881-4443
>
> On Sep 23, 2016, at 1:41 PM, Cory Quammen <cory.quammen at kitware.com>
> wrote:
>
> Simon,
>
> Thanks for your sleuthing and bug report. I don't believe your use case
> was known when merge request !488 was merged.
>
> TJ,
>
> Can you have a look at Simon's issue?
>
> Thanks,
> Cory
>
> On Fri, Sep 23, 2016 at 12:07 PM, Simon Praetorius <
> Simon.Praetorius at tu-dresden.de> wrote:
>
>> I have found an old issue in the mantis bug-tracker addressing this
>> topic: #15723
>>
>> The answer was: "VTK file format does not support saving several time
>> steps in a file" and a reference to the file-format description was given.
>> This description (version 4.2.) is definitely not complete, thus it can not
>> be an indicator for a support of writing timeseries or not.
>>
>> Although the file-format description does not show support for
>> timeseries, at least in older vtk versions (before 6.3) it was supported.
>> The source code of the XMLreader shows all the necessary parts to read the
>> time-series, all the corresponding attributes are interpreted. The only
>> problem is a change in the handling of duplicate names. This was done about
>> a year ago in merge request !488. Since there is no test that reads back
>> the files generated by `IO/XML/Testing/Cxx/TestXMLCInterface.c` no error
>> could be detected. The corresponding test to create the file was removed in
>> June 2016.
>>
>> Thus, the questions are:
>> - Is this feature quietly removed from VTK, or removed by accident since
>> only few people use it?
>> - Is the documentation of the file-format incomplete or the
>> implementation simply not cleaned up?
>> - Should the test for duplicate array names be rewritten in order to
>> allow timeseries again?
>> - There should be at least a comment in the file-format description that
>> duplicate array names are not allowed!
>>
>> Maybe this issue could be forwarded to the developer mailing list.
>>
>> Simon
>>
>>
>> Zitat von Simon Praetorius <Simon.Praetorius at tu-dresden.de>:
>>
>>
>> Hi,
>>>
>>> When reading a VTK XML-file that includes timesteps, the
>>> `vtkXMLDataReader:282` throws an error that multiple arrays with the same
>>> name are present in the file. In general this is a valid test and the error
>>> is correct, except for the case that multiple timesteps are written in the
>>> file. An example file can be created by the test
>>> `IO/XML/Testing/Cxx/TestXMLCInterface.c`:
>>>
>>> ```xml
>>> <Piece NumberOfPoints="8" NumberOfCells="1">
>>>   <PointData Scalars="example data">
>>>     <DataArray type="Float32" Name="example data" TimeStep="0"
>>> format="appended" RangeMin="0" RangeMax="0" offset="0" />
>>>     <DataArray type="Float32" Name="example data" TimeStep="1"
>>> format="appended" RangeMin="1" RangeMax="1" offset="240" />
>>>     <!-- ... -->
>>>   </PointData>
>>> </Piece>
>>> ```
>>> Here all DataArrays refer to the same dataset at different timesteps.
>>>
>>> Reading back this file, using a `VTLXMLUnstructuredGridReader` (e.g.
>>> `Examples/IO/Cxx/DumpXMLFile.cxx`), produced the error above:
>>>
>>> ```
>>> vtkXMLDataReader.cxx, line 282: Duplicate array names.
>>> ```
>>>
>>> Writing TimeSteps seems to be supported by the XMLReader and the
>>> XMLWriter. Maybe the `vtkXMLWriterC` is not valid/compatible any more? Is
>>> the structure of the vtk format changed when writing timeseries, compared
>>> to the xml-code shown above?
>>> (I'm referring to the VTK version 7.0.0, when naming line-numbers.)
>>>
>>>
>>> This issue refers to an old ParaView issue #5542 (
>>> http://www.paraview.org/Bug/view.php?id=5542). The file was readable
>>> with at least vtk-5.2 and paraview-3.6.2.
>>>
>>> Simon
>>>
>>> --
>>> Dr. Simon Praetorius
>>> Institut für Wissenschaftliches Rechnen
>>> Fachrichtung Mathematik
>>> Technische Universität Dresden
>>> Tel.: TUD-42377
>>> Mail: simon.praetorius at tu-dresden.de
>>> Web: www.math.tu-dresden.de/~spraetor
>>>
>>
>>
>>
>> --
>> Dr. Simon Praetorius
>> Institut für Wissenschaftliches Rechnen
>> Fachrichtung Mathematik
>> Technische Universität Dresden
>> Tel.: TUD-42377
>> Mail: simon.praetorius at tu-dresden.de
>> Web: www.math.tu-dresden.de/~spraetor
>>
>> _______________________________________________
>> Powered by www.kitware.com
>>
>> Visit other Kitware open-source projects at
>> http://www.kitware.com/opensource/opensource.html
>>
>> Please keep messages on-topic and check the VTK FAQ at:
>> http://www.vtk.org/Wiki/VTK_FAQ
>>
>> Search the list archives at: http://markmail.org/search/?q=vtkusers
>>
>> Follow this link to subscribe/unsubscribe:
>> http://public.kitware.com/mailman/listinfo/vtkusers
>>
>>
>
>
> --
> Cory Quammen
> Staff R&D Engineer
> Kitware, Inc.
>
>
>

-- 
Cory Quammen
Staff R&D Engineer
Kitware, Inc.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/vtkusers/attachments/20160926/a13ef962/attachment.html>