Pulumi without variables

__main__.py

import pulumi
import pulumi_aws as aws
from typing import Optional, Dict, Tuple
import asyncio

all_resources: Dict[Tuple[str, str], pulumi.Resource] = {}

def transform(args: pulumi.ResourceTransformationArgs) -> Optional[pulumi.ResourceTransformationResult]:
    all_resources[(args.name, args.type_)] = args.resource
    return None # do not actually modify the resource

pulumi.runtime.register_stack_transformation(transform)

async def get_resource(type: str, name: str, prop: str):
    while True:
        res = all_resources.get((name, type))
        if res is not None:
            return res.__dict__[prop]
        await asyncio.sleep(1000)

# User code
# -------------------------------------------------------------------------------
       
aws.s3.BucketObject("o", bucket=get_resource("aws:s3/bucket:Bucket", "b", "id"))
aws.s3.Bucket("b")

Quite a few things that can be improved and cleaned up here. But this works as a PoC for using Pulumi without variables, and as a result, allowing out-of-order resource declarations.

thank you for this! If I want to understand a bit more about what's going on under the hood there (specifically pulumi.runtime and what a transformation does and why we need def transform), where would I go?

I am struggling to find documentation on pulumi.runtime. I was able to find https://www.pulumi.com/docs/reference/pkg/nodejs/pulumi/pulumi/runtime/#registerStackTransformation ... but (a) this is a NodeJS-specific doc and (b) even the NodeJs doc is short on concepts and explanation.

There are docs on transformations and specifically stack transformations used here at https://www.pulumi.com/docs/concepts/options/transformations/#stack-transformations.

This is a slight abuse of that feature, since we aren’t actually using it to transform the resource. But it’s a tool that is available today that can enable intercepting all resource construction calls.