This message was deleted.

This is most likely not a bug but you're most likely using an

Output<string>

in a place where only

string

is accepted, which will cause the

Output<string>

to be rendered as an error message telling you that you're calling

toString

on an unresolved

Output

, which is likely to be > 255 characters. If the place where you use only accepts

string

and not

Output<string>

, you need to call

apply

on the

Output<string>

to perform the action on the

string

that the

Output<string>

eventually resolves to asynchronously, e.g.:

const imageOutput: Output<string> = ...
imageOutput.apply((image: string) => {
  // do something with image, which is the string the imageOutput resolves to
})

Hope that helps!

I know this is old, but… IMO bug or not, this is a real usability issue. If you cannot reliably produce a normal string from an Output before its value is used (me, I am using it as part of the

image

name in a task definition file, and it just flatly refuses to accept the ecr repository’s

.repository_url

, even if I try to use

apply

on it), what good is it?

…you can do this. You’re likely using the

apply

wrong

ok based on https://www.pulumi.com/docs/intro/concepts/inputs-outputs/, I’ve been doing the apply directly on the output:

repository = aws.ecr.Repository(...)

task_definition = aws.ecs.TaskDefinition(...
  container_definitions = json.dumps(
    [
      {
        'image': repository.repository_url.apply(lambda v: f'{ v }'),
        ...
      }
    ]
  ),
  ...
)

which gives me

TypeError: Object of type Output is not JSON serializable

You need to do it outside the json.dumps call

I see it and am trying it out now. Honestly I would never really have thought about doing it this way