GitLab

ENH: Update paraview3 to use xdmfnew. Fix serial (non MPI) compilation, copy xdmfwriter from xdmf1, add removegrids() method to xdmfreader to synch grids and arrays when domain changes, and protect against two uses of an null pointer for undo/redo.

number of groups is number of first level grids
datsets are the leafs of the trees.

ENH: Updated all filters to new pipeline. Updated composite filters to use DATA_OBJECT() instead of COMPOSITE_DATA_SET(). Removed part filters, multi-block datasets should be used now. Fixed a few minor bugs

BUG:RequestDataObject should create the hierarchical dataset but not create the subblocks because now the composite pipeline call PrepareForNewData in ExecuteDataStart which wipes out the subblocks

BUG: removing a debugging statement that was causing the XdmfRad tests to crash because cout doesn't know what to do with NULL

BUG: Moved creation of output to REQUEST_DATA_OBJECT. An algorithm should not create and set it's output in the constructor. This makes changing executives later impossible

BUG: get Xdmf reader to work correctly in PV using the new pipeline; changing which arrays to read is correctly propagated to the GUI