[Paraview] Exposing Timesteps in a Custom File Format

[Mike Jackson]

OK. I figured out the second question.. I just did a
outputVector->ShallowCopy(...) and that seems to work.

My first question is still out there though.. What I have figured out is
that if I "animate" then the slider is properly updated with the values that
it should have. Although it now moves in increments of 5 instead of 1..
Which makes sense as I have 500 timesteps, and this is a Scale widget.. Have
to dig deeper to see how the Exodus reader is getting away with this.

Any help is still appreciated.
Mike


On 11/8/06 5:54 PM, "Mike Jackson" <mike.jackson at imts.us> wrote:

> Thanks for the info.. I managed to get parts of my reader working but I am
> having trouble getting some others working.
> 
> The Range for Timesteps that show up in ParaView is always 0 to 100. I have
> 500 in my file. I have the public set/get methods for TimeStepRange. Is
> there somewhere else that I should have some accessors? Looking through the
> Exodus Reader is not really yielding anything.
> 
> I can post my xml files if needed.. I am sure I am just missing something
> simple.
> 
> Also, when I get a pointer to the vtkUnstructuredGrid from the OUtputVector
> in RequestData method there seems to be something fishy about using that
> Object as it causes an EXEC_BAD_ACCESS when trying to add a new cell to it.
> Is there some sort of initialization that I need to be doing? I do create my
> own vtkPoints object and set that into the vtkUnstructuredGrid? Is that a
> problem? Should I be using the vtkPOints from the grid?
> 
> Thanks for the help.
> Mike Jackson
> 
> 
> On 11/8/06 2:16 PM, "Berk Geveci" <berk.geveci at kitware.com> wrote:
> 
>> Hi Mike,
>> 
>> First a disclaimer: the time support in VTK and ParaView is evolving.
>> 
>> Currently, the only way to set the time on a reader in ParaView is
>> though a set/get method pair on the public API. Exodus does this by
>> providing SetTimeStep and GetTimeStep methods. To add the time
>> support on the GUI, readers.xml and Readers.xml files have to updated.
>> 
>> In VTK, a time aware pipeline can send UPDATE_TIME_INDEX() to request
>> a specific time step instead of using the public methods. This is not
>> yet supported in paraview. It is however on our roadmap.
>> 
>> A reader tells the pipeline (and paraview) which time steps are
>> available by setting TIME_STEPS() key on the output information during
>> RequestInformation(). This is also what is used to automatically display
>> the current time value as annotation.
>> 
>> Also, there has been some recent changes to the time support in VTK.
>> UPDATE_TIME_INDEX() is no longer used. UPDATE_TIME_STEPS() is used
>> instead. This is not index based. It contains the time values that the
>> pipeline is requesting. It is up to the reader to find the closest time
>> step or even interpolate if desired. For more information, look here:
>> http://www.vtk.org/Wiki/VTK/Time_Support
>> 
>> I hope that there will be an article on this in the next Kitware Source.
>> 
>> -berk
>> 
>> On 11/8/06, Mike Jackson <mike.jackson at imts.us> wrote:
>>> I am developing a ParaView reader for my custom file format. Inside our
>> data
>>> we have the data laid by timesteps. I was wanting to use this information
>>> inside of ParaView so I am basically copying the Exodus Reader (at least
>> the
>>> timestep portion). So far everything is ok except that the following line
>> of
>>> code comes back false:
>>> 
>>>   if(outInfo->Has(vtkStreamingDemandDrivenPipeline::UPDATE_TIME_INDEX()))
>>> 
>>> It is with in this block that the time value and the index get updated but
>>> for some reason when run in my code this comes back false. I am currently
>>> running the reader using a command line "test" program and have not
>> actually
>>> tried it in ParaView for some other reasons. Is there a special type of
>>> pipeline I need to instantiate in a non-ParaView setting? Is there
>> something
>>> is the xml files that needs to be set so the UPDATE_TIME_INDEX key is
>>> present? How about in the header file for my class? Maybe some
>> undocumented
>>> accessor method?
>>> 
>>> Any help would be appreciated.
>>> 
>>> --
>>> Mike Jackson   Senior Research Engineer
>>> Innovative Management & Technology Services
>>> 
>>> 
>>>