ayushpandey8439 / distflow Goto Github PK

The logic written in sampleRun.erl can be moved into runner.erl and static functions can be made for execution of individual tasks which take care of extracting the data (input and output) themselves.

Erlang message passing has backpressure. Replace the message passing with message queues.
ATM, there are no gurantees that the task runner gives in terms of order of execution. Add gurantees in the distribution tasks.

Make the runner a normal erlang module instead of a genstatem andtracted module.

Think about how to pass node outputs from one node, not just to the next but also named inputs as specified in the specification.

The current map operation calls the functions directly. This should be replaced with a dispatch operation which will make the execution of long running tasks async.