Serverless Stack

Okay, I have two questions about sst start : Is the stack created based on me ? what if another dev starts the project for local dev while I am already developing from local ?

this is something I also have to better understand, and similar questions have been posted recently too.

I personally believe that having a commit-eable `sst.json` is not a good idea and _maybe_ we should move to a more flexible approach based on environment variables (like .env)

<@U01MV4U2EV9> mentioned above that they use an different AWS account for each developer but I am pretty sure that this is not what everybody is doing… for example, we usually prefix our CDK stacks with our github username when developing locally – but it’s always the same “dev” AWS account

(also because if you have even just a few services, it could not be super simple to provision each ‘personal’ developer’s AWS account)

If you're not using an aws account per dev I'd recommend using a different stage per dev.  You can remove stage from `sst.json` and set it via an environment variable in the cdk entrypoint

good to know! It’s not clear (to me) how “deep” are the sst.json vars integrated in sst

Actually even simpler is just doing `sst start --stage $STAGE`

and just making sure devs have that set locally

we derive a lot of datas from ssm prefixed with the stage

so the stage determines the env variables of our lambdas

so the develop stage points to all our develop resources

Is there a reason that has to be derived from `stage` and not another environment variable?

actually no, not really, we could use some env var and keep stage for other means

I'll try that once I can start the project (stuck with wrong typechecking - cf my next comment)

We need to a better job documenting this. So debug stack is created with the stack name `$stageName-$appName-debug-stack`

So if two ppl are both running `sst start` for the same app with the same stage and deploying to the same AWS account, it’s going to conflict.

By conflict i mean whoever connects the last will receive the request.

And the person who connected first will get disconnected with a message saying another client is connected.