[Insight-users] how can I tell the pipeline that my filter needs to be re-executed?

[Neuhaus Jochen]

Hi all,

I wrote my own filter class (derived from itk::ProcessObject) that
processes my own data objects (derived from itk::DataObject). My filter
is a source, it does not have any other data objects as input. 
When I call myFilter->GetOutput()->Update() the first time,
GenerateData() is executed, as expected. If I call it a second time, the
update mechanism thinks that everything is unchanged and doesn't call
GenerateData() again. But in my special case, I need my filter to always
be modified and therefore always execute GenerateData() again. There is
no explicit state change in my filter objects, it is more that the time
has changed on the second call and GenerateData() will produce a
different output at a later time.

I'm trying to understand the pipeline update process, but I'm a bit lost
to which method I should change. Any hints for me? 

Best regards, 
 Jochen Neuhaus
---
Jochen Neuhaus
Deutsches Krebsforschungszentrum
Medizinische und Biologische Informatik
Im Neuenheimer Feld 280
D-69120 Heidelberg
Tel.: +49 (0) 6221 / 42-2329
Fax: +49 (0) 6221 / 42-2345
http://www.dkfz-heidelberg.de/de/mbi/people/Jochen_Neuhaus.html