Bring your towel and join one of the fastest growing data communities. Welcome to our second-generation open source orchestration platform, a completely rethought approach to dataflow automation.

Prefect Community

Hi folks, I'm running into an issue when trying to register a flow which has some labels already added to the run config. Calling `flow.register()` on such a flow fails with error `'list' object has no attribute 'update'.` A reproducible example is in the thread. Has anybody come across this before or knows how to resolve it?

Here's the code:
```from prefect import Flow, task
from prefect.run_configs import UniversalRun

@task()
def hello_world():
    print("hello world")

with Flow("hello_world") as flow:
    hello = hello_world()

flow.run_config = UniversalRun()
flow.run_config.labels = ["foo", "bar"]
flow.register(project_name="test-project")```
The stacktrace states that the problem occurs in the `obj.labels.update(self.storage.labels)` line of this section of `prefect.core.flow`:
```# add auto-labels for various types of storage
for obj in [self.environment, self.run_config]:
    if obj is not None:
        obj.labels.update(self.storage.labels)
        obj.labels.update(labels or ())```
If I remove the `flow.run_config.labels = ["foo", "bar"]` line in my example code, the flow registers successfully.

Hi <@U01AGSHH6UR> - I'm going to open a ticket in the Core repo for this; in the meantime, try putting your labels on the register call (`flow.register(labels=["foo", "bar"])`)

<@ULVA73B9P> open "Flows with run config defined in code fail the registration step if labels are defined"

<https://github.com/PrefectHQ/prefect/issues/4300>

Thanks for the super-speedy reply :slightly_smiling_face:

The workaround does allow the flow registration to succeed, thanks for that!

This is because `run_config.labels` is a set, not a list. When you manually assign things like this (rather than passing the labels to the run-config construtor) you're using the wrong type and breaking the internals.

Try:

```flow.run_config = UniversalRun(labels=["foo", "bar"])```

Got it, thanks a lot <@U011EKN35PT> - the example I posted was part of a bigger "flow preparation" routine we're using, where the run config was being instantiated first and then different labels were added to it afterwards depending on the use case of the flow. Changing that logic to first work out all the labels to apply, then instantiating the run config object, has fixed the root cause. Thanks again :slightly_smiling_face: