Arrays in parallel udf
 

hello,
I have a doubt about arrays defined in parallel udf, plz help me.
Below is some of my opinion, but i don't know if that is right or not.

If I defined an array in a parallel udf without using compiling directives such as #if RP_NODE, the host and all the other compute nodes would have an independent copy of the array with the same name. If a node modified some elements of the array, the other nodes would not automatically see the modification.

Am i right?

In my opinion unless you use RP_NODE or RP_HOST the changes in array would not be communicated between nodes and host.