[vtk-developers] Recent changes to vtkCompositeDataPipeline

[Bill McGrory]

I had previously reported a kludge which allowed for a particular
pipeline consisting of composite data and simple filters.

After a recent CVS update, my kludges are no longer relevant, but my
pipeline no longer works either.

I believe that some of  my issues with proper choice of composite data
and simple data within the pipeline have been resolved (In other words,
I think the recent updates have fixed some of the problems that I was
working around), but I still get errors related to invalid update
extents.

To refresh,

I have a number of derived vtkStructuredGridAlgorithm's feeding into a
single vtkMultiGroupDataGroup filter. This filter feeds into a simple
filter (let's say for purpose of discussion, a vtkStructuredGridOutline
filter. This feeds into a vtkMultiGroupDataGeometry filter, and to a
mapper.


The problem that I see, is this.

Within vtkCompositeDataPipeline::ExecuteSimpleAlgorithm

The call to REQUEST_UPDATE_EXTENT is filled with a whole extent of the
last structured grid in the multigroupdata object. So the
outputInformation(WHOLE_EXTENT()) for the structuredGridOutline filter
does not get set correctly to the extent of the structuredgriddata
associated with INDEX().


If I look at the pipeline behavior, the ExecuteInformation call
associated with the structuredgridoutline filter within the simple loop 
does not propagate upstream to get a proper Extent for the current
INDEX.

I know that this is where the problem ends, and results in an error, but
I'm not certain how to fix it. Can I send a call which makes the
requestinformation pipeline time invalid, and make it propagate back to
the current structuredgrid algorithm?

Thanks
And regards
Bill