Flyte enables production-grade orchestration for machine learning workflows and data processing created to accelerate local workflows to production.

Flyte

<@U01J90KBSU9> in the video you linked, they briefly say you could implement "periodic triggers" using `sleep` (attached screenshot). But there is no example given, unfortunately.

Therefore, I'm asking here again: How would you use `sleep`  to trigger a task periodically inside a `@dynamic`workflow and how would the DAG look like in that case?

Here's an example:
```from datetime import timedelta

from flytekit import dynamic, task
from flytekit.core.gate import sleep


@task
def t1(a: int):
    ...


@dynamic
def wf_sleep(n: int):
    for i in range(n):
        x = sleep(timedelta(minutes=10))
        b = t1(a=5)
        x &gt;&gt; b```
`t1` task will be triggered `n` number of times with a fixed interval of 10 minutes between every two triggers.

My misunderstandings come from the fact that I expected `@dynamic` workflows to be "real" dynamic. But unfortunately, they still require a known DAG. "Real" dynamic DAGs would be a nice addition to flyte imho

```from datetime import timedelta

from flytekit import dynamic, task
from flytekit.core.gate import sleep


@task
def t1(a: int):
    ...


@dynamic
def wf_sleep():
    while True:
        x = sleep(timedelta(minutes=10))
        b = t1(a=5)
        x &gt;&gt; b```

i.e. not just DAG support but also loop support

above `dynamic` workflow cannot be evaluated pre runtime (while True)

but as I understand it, this is a requirement

what I need is truly event-driven task pipelines I guess

Yeah. `while True` won't work. There should exist some sort of determinism.

Hi <@U04N000P37W> can you expand on what you mean by “real dynamic”? Might be useful to write psuedo-code in plain Python to show the desired behavior

<@U01DYLVUNJE> his code indicates that he wants to interact with the outputs of a task execution in python and then launch another task. This is what we have been calling eager or async/await