In case you missed my first post “Implementing a design system”, we chose to implement our design system with web components. This post will be about our thought process and the tradeoffs we considered when deciding on how we would implement with web components.

It is possible to write a web component using only native JavaScript and HTML, but it is challenging. It is much more common to use a library like Lit or Stencil. The library helps with registration of the web component and change management as well as interoperability with popular frameworks. In our case, in addition to Lit and Stencil, we also considered Mitosis, and compiling to web components with Solid or Angular.

In the Devies book club we recently finished Modern Software Engineering by David Farley.

In Modern Software Engineering, Farley want to reclaim the meaning of engineering in “software engineering”. It’s not just about coding, it’s about doing what works to build better software faster - This involves processes, tools and culture.

The software industry is not like the manufacturing. The process of building software can’t be reduced to a production line; it’s rather a process of exploration and discovery. In this context, Farley suggests that there are two critical core competencies that software engineers must master: learning and managing complexity.

Sometimes I see people (myself included) working on software development tasks in a non-structured way. You might have a vague understanding of what you’re about to accomplish. Aimlessly traversing the code base, hoping to find places that you might need to change. You instantly start to chase new ideas, while forgetting about prior (and possibly important) ideas. It might be hard to determine when you’re done. You might not commit intermediary working states, and have a hard time finding what you’ve messed up once something is broken.

Background

This is probably not news to anyone that has been working as developer consultant. Large existing code base; in our case it’s an application that has existed for about 2,5 years and during that time it has developed several services and therefore has many endpoints.

Therefore when me and my colleagues had to start developing new features it was hard to know what was available and what wasn’t with over 50 different endpoints to hit.