context manager capabilities for cluster config about pyjaws HOT 5 CLOSED

Hey @rafaelpierre, this is for syntactic sugar? We don't seem to access any resources in the cluster.

Adding something like this to Cluster...

class Cluster(BaseModel):
   ...
    def __enter__(self) -> Cluster:
        return self

    def __exit__(self, *args):
        pass

We can use it like this...

with Cluster(...) as cluster:
    task_1 = NotebookTask(
    ...
    cluster=cluster)

I've seen with syntax used quite differently in the kubeflow dsl though, ExitHandler & ParallelFor for example.

from pyjaws.

Yes, exactly @aaravind100 - this is more syntactic sugar :)

we should keep both flavors available: 1. assigning clusters to tasks (which basically maps the cluster_key to the task payload) and 2. with the context manager - with Cluster(...) as cluster as you pointed out

wanna take a stab at it?

from pyjaws.

Yep, I can take this. I'll create a pr in sometime.

from pyjaws.

Amazing @aaravind100!

from pyjaws.

Closed with #14

from pyjaws.