My understanding of the code here is that AWS_SESSION_TOKEN, AWS_SECRET_ACCESS_KEY, and AWS_ACCESS_KEY_ID are assigned to environment variables, and later inferred by the S3 client and the DynamoDB client. It would be nicer if they could be passed directly to these clients without being part of the process's environment.
I think an ideal solution would allow me to configure my own object store and dynamo client, and provide those when registering a table.
Use Case
My process is a big monolith with a bunch of stuff in it, completely unrelated to me. Many different identities are used in this process for many different purposes. Environment variables are a single global namespace, so it would be dangerous to plug my credentials in it.
Description
My understanding of the code here is that AWS_SESSION_TOKEN, AWS_SECRET_ACCESS_KEY, and AWS_ACCESS_KEY_ID are assigned to environment variables, and later inferred by the S3 client and the DynamoDB client. It would be nicer if they could be passed directly to these clients without being part of the process's environment.
I think an ideal solution would allow me to configure my own object store and dynamo client, and provide those when registering a table.
Use Case
My process is a big monolith with a bunch of stuff in it, completely unrelated to me. Many different identities are used in this process for many different purposes. Environment variables are a single global namespace, so it would be dangerous to plug my credentials in it.
Related Issue(s)