Hello guys! Question! How can I reference one stack in another stack? In your example, you create DynamoDB Stack + the API in the same file. But what if I want to create a separate file, one for DynamoDB and one for the API, then allow the API to access the DynamoDB(my table) in my API file.

You can pass stack's

this

object. It's code, so passing around references are fine.

I wondered if it’s the best practice. This what I did for information. Thx @Dennis Dang DynamoStack; 

export default class DynamoDB extends sst.Stack {

 

constructor(scope, id, props) {

  

super(scope, id, props);

  

this.table = new sst.Table(this, "xxx", {

   

fields: {

    

PK: sst.TableFieldType.STRING,

    

SK: sst.TableFieldType.STRING,

   

},

   

primaryIndex: { partitionKey: "PK", sortKey: "SK" },

  

});

 

}

}

MyApi; 

export default class MyApi extends sst.Stack {

 

constructor(scope, id, props) {

  

super(scope, id, props);

  

const { table } = props;

  

const api = new sst.Api(this, "Api", {

   

routes: {

    

"POST /projects": "src/project/create.main",

   

},

  

});

  

api.attachPermissions([table]);

 

}

}

App file

export default function main(app) {

 

const { table } = new DynamoDB(app, "dynamodb");

 

new Cognito(app, "cognito-auth-service");

 

new ProjectAPI(app, "api-project-service", { table });

}

That looks good to me. I believe CDK is still so new that there's no best practice. Even if you peruse all the cdk examples from the aws-examples repo, they're almost all very simple, and provide little abstraction to help organize code. But in principle, we're mutating this giant stack object that gets transformed into cloudformation.

@Bma that’s what I would do too. In this case CDK auto-creates a stack output of the table’s ARN (b/c that’s what’s being used by the attachPermissions call in the API stack) in the DynamoDB stack and the exports it. And the API stack auto-imports the it.