what's wrong with this ?

You need to unlearn the syntax of the language you are used to. You create a variable and assign a value in Appian just using "=" ":" not ":=".

And, Appian uses a functional language paradigm where variables are immutable. So your foreach will also not work.

and one more thing stefan can you please guide me here

I want my result to be in a single list {a,b,c,d,e,f} but I am getting result as  {{a,b}, {c,d}, {e,f}}

You can try using a!flatten.

You try to load all data into memory. That will not work. Appian has some low and hard limits on resource allocation. What are you trying to achieve?

And to learn the basics of Appian, I highly recommend the Appian academy, tutorials and documentation. And I am running a blog and podcast on https://appian.rocks.

I have mentioned it earlier as well that I want to compare data from appian report and the data present in database, I know you have mentioned about the procedure also but that's not possible as I need those unique details further as well which makes my solution more complicate and here I am fetching the data in batches and then using it , now is that also a problem in appian ?

Depending on the data volume, this will get you in trouble. So I need to repeat my suggestion from the other thread: Write the data to a DB table and run a stored procedure.

considering there is a purging also scheduled.. may I know how much volume it can handle properly without any trouble so I can schedule the purging accordingly.. 

docs.appian.com/.../Memory_Circuit_Breaker.html

This is a super confusing way to construct this for the end result you're apparently after.

If you want to try to query all data in the data store, you should just do that in your initial query - and pass in a Query Selection value to only pull back the column you're interested in, which will make the query more efficient (instead of what you're currently doing, which is to apparently force the query to grab all columns, then using index() [instead of property(), grumble] to grab the "processId" property from the result, meaning you're consuming more query bandwidth than necessary.

This is a super confusing way to construct this for the end result you're apparently after.

If you want to try to query all data in the data store, you should just do that in your initial query - and pass in a Query Selection value to only pull back the column you're interested in, which will make the query more efficient (instead of what you're currently doing, which is to apparently force the query to grab all columns, then using index() [instead of property(), grumble] to grab the "processId" property from the result, meaning you're consuming more query bandwidth than necessary.