pablopunk/iga

Zero config typescript/es6 server with the filesystem as the router

iga's Issues

in-memory cache

• One for request.url with {body, code}
• One for endpoint handlers (like the one that's on master but with a new module)

This cache will improve requests/sec by 10x based on my experiments

Export a handler instead of creating the server

Make src/index.js export the handler, no need to create the server in that library. This way I can use it as a middleware with any server, while keeping the ability to create a server inside src/cli.js.

This will allow easier deployments, for example compatibility with now 2.0 lambdas.

NOTE: This means all manual cache mechanisms should be implemented where the server is implemented i.e the CLI. Http cache will still be managed in the library (manually) but for example handlers cache should be moved to the CLI. This affects #6

code reloading

iga command should do this

typescript

Use typescript with sucrase if the file is ts

Extend response and request objects

Right now both are the standard NodeJS http response and request objects. It's okay because people know how to use them and they're very well documented, but I would like to add a bit more functionality to both objects.

Some ideas:

• request.query: An object using require('url').parse
• reponse.json: Sets headers for JSON and accepts an object (that will get stringified before sending)
• return object | string | number: Instead of using the response object, it would be usefull to just return a value. An object should be sent as a json, a string as text, and the number could just be the status code.

Example:

// find-user.js
export default (req, res) => {
  if (!req.query.id) {
    return 403
  }

  const id = req.query.id
  const user = getUserById(id)

  if (!user) {
    return 404
  }

  return { items: [user] }
  // or res.json({ items: [user] })
}

• json
• string
• query params
• status code

Consider replacing deprecated `url.parse` with `url.URLSearchParams`

https://nodejs.org/api/url.html#url_constructor_new_urlsearchparams_string

This would be a breaking change as the returned values need to be accessed by get and set, and right now it's just a plain object.

Replace `nodemon` with a lighter alternative

Nodemon is the dependency that makes iga so heavy in size

Consider using chokidar https://codeburst.io/dont-use-nodemon-there-are-better-ways-fc016b50b45e