React.js (2 blogmarks)

I’ve only heard good things about TanStack (Start) and have been wanting to try it. This post may be the encouragement I need.

With Next.js, a combination of the push for the App router and the move to RSC shattered the original simplicity of React.

The app router is riddled with footguns and new APIs, unrelated to React, but sometimes blurring the line with it. It’s hard to know when Next.js begins, and React ends.

"One of the many ways this matters is through testing. Since switching away from React, I’ve noticed that much more of our application becomes reliably-testable. Our Capybara-powered system specs provide excellent integration coverage."

"When we view the lines of code as a liability, we arrive at the following operating model: What is the least amount of code we can write and maintain to deliver value to customers?"

Not all lines of code are equal, some cost more than others to write and to maintain ("carrying cost"). Some have a higher regression risk over time than others.

"When thinking about the carrying cost of different lines of code, maintaining different levels of robust tests reduces the maintenance fees I must pay. So, increasing my more-difficult-to-test lines of code is more expensive than increasing my easier-to-test lines of code."

Language, in as much as it relates to testability, is the metric of focus here. What other facets of code increase or decrease their "carrying cost"?