ForEach Task

Iterator task will go through an input list or dictionary and will trigger tasks connected to the yellow port.

Each iteration will provide the following variables:

• loop_index
• loop_key (dictionary only)
• loop_value
• loop_size
• end_of_loop (True/False)

Input
Python list or dictionary. If a string is provided, it will be decoded as a json component.

Iterate if
Triggers the yellow port only if condition is verified. This parameter is evaluated before each iteration, and loop_index is incremented even if iteration is not triggered.

Pseudo code equivalent

    For_each value in INPUT  

      If ITERATE_IF is True
        Execute Yellow port
      Else
        Next value
        loop_index = loop_index + 1

Max running
Defines the maximum concurrent iterations. If greater than 1, iterator will trigger a batch of n iterations and wait for termination of the first one before triggering another one. Note that an iteration could be composed of various links, like the following example.

In this case, 2 iterations will be triggered at the same time:

Iteration loop_index 0:
- trigger task A
- trigger task B

Iteration loop_index 1:
- trigger task A
- trigger task B

Once an iteration is terminated (i.e tasks A and B are ended), a next one is triggered to reach max running iterations.

NB: there is no guaranty about the tasks order when iterations are triggered (task A could be started after task B).

Iterate for each link

Instead of triggering all linked tasks for 1 iteration, this parameter will trigger 1 iteration by linked task. Each task will receive a unique loop_index, and a new iteration will be triggered as soon as a previous one is terminated.

The previous example will generate the following output.

• Tasks A and B are first started, with respectively 1 and 0 as loop_index (note that task B started before task A, independently of the graph representation).

• While A is sleeping for 20 seconds, B will be executed 3 more times, as it’s sleeping only for 5 seconds. Each of started tasks B will have a different "loop_index".

• And so on, depending of each task duration….

Batch size
If greater than 1, each loop_value will be a subset of n input elements.

This example will produce 4 iterations with the following loop_value.

Note:

• loop_index is relative to the subset, and not to the global input.
• The final subset could have less than batch size elements.

Sleep
Defines the amount of execution sleep time between each iteration.