Tom MacWright

React is old

My last big project at Mapbox was working on Mapbox Studio. We launched it in 2015.

For the web stack, we considered a few other options - we had used d3 to build iD, which worked out great but we were practically the only people in the internet using d3 to build HTML UIs - I wrote about this, in “D3 for HTML” in 2013. The predecessor to Mapbox Studio, TileMill, was written with Backbone, which was cool but is almost never used today. So, anyway, we decided on React when we started it, which was around 2014.

So it’s been a decade. Mapbox Studio is still around, still on React, albeit many refactors later. If I were to build something new like it, I’d be tempted to use Svelte or Solid, but React would definitely be in the running.

Which is to say: wow, this is not churn. What is the opposite of churn? Starting a codebase and still having the same tech be dominant ten years later? Stability? For all of words spilled about trend-chasing and the people talking about how one of these days, React will be out of style, well… it’s been in style for longer than a two-term president.

When I make tech decisions, the rubric is usually that if it lasts three years, then it’s a roaring success. I know, I’d love for the world to be different, but this is what the industry is and a lot of decisions don’t last a year. A decision that seems reasonable ten years later? Pretty good.

Anyway, maybe next year or the year after there’ll be a real React successor. I welcome it. React has had a good, very long, run.