annesadleir / rjr-aws-lambda-base

Implements the API for an AWS Lambda runtime

rjr-aws-lambda-base

Implements the API for a lambda layer: polls for input text, processes it, and posts back the result. It has the concept of a LambdaWorker, which is the class that does the work specific to that lambda, analogous to the RequestHandler when using the standard AWS java runtime.

This is just the runtime part. To see how to use it, look at the rjr-aws-lambda-dummy repo which has an example with a piece of code that just repeats Strings. (That repo's readme also has deployment instructions.)

This code is not intended as a general-purpose layer which can be deployed separately as a dynamic base for different lambdas. Instead it is always intended to be used as a dependency of a specific lambda's code, and compiled into a deployment package with it. Therefore it can avoid using reflection to instantiate the LambdaWorker at runtime.

There is a reflect-config.json file in META-INF/native-image for use with GraalVM native compilation.

I used this guide as my starting point, and my code is heavily based on it (though it's not in Kotlin):
'Fighting cold startup issues for your Kotlin Lambda with GraalVM' by Mathias Düsterhöft
I also used the AWS custom runtime API definition:
AWS Lambda Runtime Interface
which includes a link to a definition of the API in OpenAPI/Swagger format. I also used
'How to Deploy Java Application with Docker and GraalVM' by Vladimír Oraný
and various other web resources like the GraalVM docs and other AWS docs. And lots of StackOverflow, obvs.

todos

proper logging rather than System.out.println()?