serverless-reasonml

This is a serverless framework plugin that has been written for personal exploration while learning ReasonML and being curious about its plausible usage writing serverless functions.

It has been intentionally made to be compatible with the serverless-offline plugin.

Compared to great plugins like serverless-webpack and serverless-offline, this plugin is extremely trivial. Therefore assume there lots of edge cases that isn't covered yet, although it has been valuable for personal use to get builds triggered at the right time.

Usage

$ npm install --save serverless-reasonml bs-platform

Create a bsconfig.json with bucklescript configuration. Somewhat sensible example:

{
  "name": "using-serverless-reasonml",
  "version": "0.1.0",
  "sources": {
    "dir" : "src",
    "subdirs" : true
  },
  "package-specs": {
    "module": "commonjs",
    "in-source": true
  },
  "suffix": ".js",
  "warnings": {
    "error" : "+101"
  },
  "namespace": true,
  "refmt": 3
}

The example above expects source code to be put in the src directory.

Finishing it off, the serverless framework has to be told to use this plugin by putting the following in serverless.yml:

plugins:
  - serverless-reasonml

Known challenges

bs-platform is required in production because it contains helper functions used by the code bucklescript generates. At the time of writing this, there's lots of development tools included in bs-platform which results in a huge .zip file being created by the serverless framework.

The initial experiements I've done so far, results in archives near 50MB for extremely simple functions.

There are ongoing efforts in bs-platform splitting those development tools from the production utilities, until then, the archives created with this setup will be big in filesize. That's quickly a showstopper for Lamda@Edge functions which has a lot lower limit in regards to function archive filesize.

Related resources

• @ahrefs/bs-aws-lambda for extensible AWS Lambda typings
• serverless-offline to run functions trigger by HTTP requests locally