How is everyone handling running Typescript async functions in their sst app? I'm retrieving some data from DynamoDB that's required to know what to build in each stack. I think the gap in my current approach is the best way to provide the async data to the functional stack. Right now I put an init function which I call from the sst App index.ts module which will run the async process. Each stack init is calling the same module which caches the data from DynamoDB. Ideally I could retrieve the info in the top level App index.ts and inject it into the stacks that need it. I'm just not seeing a way to do it. Here's an example of my current approach which won't scale very well:

export default async function main(app: App) {
	await stack1Init();
	await stack2Init();
	app.stack(stack1)
	.stack(stack2)
	.stack(stack3)
}

I could also be doing something that wasn't intended and I'm looking for feedback on that as well. I'm looking for a better way to do this.

Are you saying `await app.stack(func)`is not working?

It's not. It ends up being void

I actually haven't confirmed this myself but try this

const MyAsyncData = await something()

function MyStack() {
  MyAsyncData
}

export default function main() {
  app.stack(MyStack)
}

Yeah, not sure why I didn't think to try that. I'll do that and let you know

but to answer your other question the way to inject data is to use functions that return functions

function MyStack(args) {
  return function MyStack(ctx) {
    ...args
  }
}

app.stack(MyStack(args))

If it helps I'm on version 1.2.13

Probably not helpful, but one thing to mention is that having your stack dependent on external data is considered an anti-pattern in CDK. People in the CDK world would strongly encourage you to not do this. The argument being that your CDK should be deterministic, so instead of a dynamodb call you should have a local config file. Not sure how people in the SST world feel about this

Thanks for pointing this out and that would have been my preference if it was possible. My use case doesn't lend itself to local configuration data though. I'm deploying a stack automatically for a certain tier of user. So this is running dynamically in a CodePipeline started by an API request. To provide it statically I could probably export the data to s3 or an artifact, but the DynamoDB solution will work for now. I'm open to other suggestions.

So we are doing something similar. Instead what we are doing is when a new user is created our database is updated. We then put the user configuration information in a gh repo. In our pipeline we pull the user configuration from the repo and deploy the infra, that way give certain user config + cdk we always get the same output. Everything is versioned controlled so we can roll-back to previous states if need be