There is a bug* in the documentation that removes the generic

Output<string>

from all

string

types, and my guess is that you're picking up on that, and defining the

scope

value to have type

string

. But it needs to be

Output<string>

. (*: Pulumi states "it's correct by design" but it confuses all newcomers, so I call that a bug.)

Solution: The solution was to change the receiving variable type for scope in my roleAssignment class to type Input<string>. To use this variable type you must import { Input } from. "@pulumi/pulumi"; Once this is done it sets up a dependency between the output value from the Vnet id and the input to the roleAssignment class. This essentially makes pulumi wait until it has a value for the input before it calls the dependency. You can achieve something similar by setting a variable by using the apply function. This code snippet prints the value as a pulumi warning then sets a variable that when you reference it pulumi will wait till the apply has run so you have a value in that variable.

pulumi.output(args.scope).apply(v => {
            pulumi.log.warn(v)
        });
         const vnetId = pulumi.output(args.scope).apply(v => {
            return `${v}`
        })

The apply function makes pulumi wait until the Output variable has a value before it executes the apply on it. This means your code will pause and wait for this to occur when you attempt to reference the variable vnetId. It's all about parallelism and making pulumi wait until an Output object has a value before you can reference it. Thanks @steep-sunset-89396 for explaining this to me. I hope my poor and slightly uninformed explanation helps anyone searching for this.

If your code needs to use the vnet id, then you would have to wait.. but in that case, your code would need to be inside the apply. You can't run the apply then do stuff afterwards. The "afterwards" code runs before the code inside the apply.

@able-engineer-79880 clearly understood how to pass outputs through the interface and the connection Input/Output wasn't so clear. Same with the apply() as a way to "reveal" the value at the Output (but not as a way to extract the value).

And finally, showing pulumi.output() was also a good exercise to convert the Input and demonstrate how to use the apply() in his code if needed.

So it was much more about common patterns than what was strictly required to actually get the code working.