Jos de Jong

tl;dr In order to ease full stack JavaScript development, we need to take away the hurdles keeping these worlds separated. We need unified, isomorphic solutions for common functionality like module loading, networking, background processes, logging, and others. Ever had to use statements like the following? if (typeof window !== 'undefined') { // browser } else { // node.js } It’s a simple check to see whether a script is being executed in in a browser, and if not, assume that it’s a node.js process. I noticed that I need them more and more. I mostly develop libraries. As soon as I need functionality which is not standard available in both browser and node.js, like fetching urls, I have to create two different implementations and switch depending on the environment. It’s annoying and I don’t yet have a seamless solution for dealing with environment specific code.

A while ago I read the blog post Ranging Near and Far by Scott Sauyet. On Echo JS, Scott titled a link to this blog “Choosing simple APIs over flexible ones”. This title suggests that you have to choose between either simple or flexible APIs, which is a false dilemma. This got me thinking: What makes a good API? Flexibility Scott discusses the API of the range function of Ramda compared to that of Underscore. Where Underscores range supports optional start (defaulting to zero), custom step size, and a negative step, Ramda’s range “only” supports a step size of 1 and does not allow to omit the start value. While Underscores implementation is more flexible, Scott argues that Ramda’s more limited implementation may be the better choice because of its simplicity. Scott suggests that it is way more complicated to use Underscores range because of it being more flexible. Scott has an argument here, but there are more factors determining how easy it is to use an API.