vscode decorator for the dynamic task

[pingsutw]

Tracking issue

NA

Why are the changes needed?

To support vscode decorator for the dynamic tasks.

What changes were proposed in this pull request?

Set the correct resolver in the current context, so flytekit can compile the vscode task

How was this patch tested?

from flytekit import task, workflow, ImageSpec, Resources, dynamic
from flytekit.interactive import vscode

new_flytekit = "git+https://github.com/flyteorg/flytekit@cedb8188f6fa86481064fb9022b16c9f5d369f4d"
image_spec = ImageSpec(registry="ghcr.io/unionai-oss",
                       packages=[new_flytekit, "kubernetes", "mypy"],
                       apt_packages=["git"])


@task(container_image=image_spec, enable_deck=True, requests=Resources(cpu="2", mem="2000Mi"))
def train():
    print("forward", flush=True)
    print("backwarddddf")


@dynamic(container_image=image_spec)
@vscode
def d1():
    print("dynamic", flush=True)
    train()


@workflow
def wf():
    train()
    d1(

Setup process

Screenshots

Check all the applicable boxes

• I updated the documentation accordingly.
• All new and existing tests passed.
• All commits are signed-off.

Related PRs

Docs link

[eapolinario]

why do we need this if the correct task resolver is being set in compile in flytekit/core/workflow.py?

[pingsutw]

I set the task resolver in the ctx.with_compilation_state(...), but there are some places where new_compilation_state() get called, potentially causing the loss of the resolver if it's not set.