Without an Assign node (in 6.4), your two options to create a DataGrid-type calculated variable which can be reused further in a diagram would be:   1. A Process Node (may be a dummy SAS process that does nothing). Its code may look like this (i did not test, but I think that should work)   package dummy_dgprocess / overwrite=yes; method execute( package tap_table DATA_GRID ); end; endpackage;   Beware though that there is a limitation while working with Data Grids in SAS DS2 code - a data grid cannot be too large to exceed 8192 characters while serialized. If it becomes that large, the diagram will fail - even though your dummy code does nothing with that data grid, it will probably still fail when SAS attempts to pass the data grid over from RTDM Engine to Fed Server and deserialize it.    There is no such limit when using Groovy language instead of DS2. But it does not have a very official support status in current versions (including 6.4): http://support.sas.com/kb/60/630.html 2. Subdiagram node to a dummy RTDM diagram that accepts a data grid input.  This option should not have a data grid size limit.   Good Luck! ... View more

You can't use data grids or arrays directly in branches or filters, since you can't compare a discrete value to a list or table. You can, however, use a calculated item against either a data grid variable or against a single column (which is represented as a list-type variable) in those nodes.   For example, you can branch based on whether a column contains a specific value by creating a boolean calculated item called "Contains XYZ?" that does "ARRAYINDEX(<<dataGridIdentifier.myColumnName>>,'XYZ')" which will return a Yes/No that you can branch on. ... View more