AWS Lambda F# Lab

Trying out F# on AWS Lambda.

Status

The build is not working correctly when compiling from Mono to .NET Core (both with build.fsx and build.sh).

The Dockerfilealso tries building an F# application with Microsoft's own dotnet in the Docker image. It is commented out, however, as the FSC compile step fails.

Manually installing the latest dotnet binaries with the F# generator and compiler does not resolve the issue (it cannot currently compile its own generated F# template projects).

The conclusion for now is to wait for a new version of the .NET Core SDK and tool-chain for Debian.

Developing with Mono on Mac

Assuming that you already have Mono installed, you need to install the Paket package manager and use it to download the dependencies.

mono .paket/paket.bootstrapper.exe
mono .paket/paket.exe install

After this, you can build the Lambda deployment package using the build.fsx script.

mono packages/FAKE/tools/Fake.exe build.fsx

Deploying the Lambda function

You can use the aws command like client like so to create the function:

aws lambda create-function \
  --function-name foo\
  --runtime dotnetcore1.0\
  --role lambda_basic_execution \
  --handler "HelloWorld::HelloWorld::handler" --zip-file "fileb://build/LambdaFunction.zip"

Use the deploy.sh script to update the function. You must set the AWS_PROFILE environment variable to the name of the AWS profile to use for deploying. (The profiles are defined in your ~/.aws/config configuration). For example if you have a profile named mjul-lambda-dev:

AWS_PROFILE=mjul-lambda-dev ./deploy.sh