Astro and release velocity
This website is built with Astro, and overall I’ve been happy with my choice. The framework model makes sense to me, and I really enjoy using their templating language for building UI. If I were to start over today, I would still choose Astro.
But I’ve had an unsettled feeling about Astro and this site, and I think I’ve finally put my finger on it: Astro is moving too quickly. It feels like every few weeks they’re putting out a new version with new features and new patterns to build new websites.
And as exciting as that is (I’m sure the maintainers are thrilled to be able to ship new things so quickly), it’s overwhelming for me as a user. I feel like I’m always behind because I am. I updated the Astro dependencies on this site two months ago, and I’m already four minor versions behind. That’s two minor versions per month!
Now full credit should go to the Astro team for working to make updates relatively painless. I haven’t encountered any particularly difficult breaking changes, and they have tooling that helps make the process simpler. But a simple process is still a process, and it’s one that users have to do a lot to keep their Astro sites up to date.
I know that no one is forcing me keep my dependencies on latest; that pressure is internal. But it’s real, and it impacts how I feel about working with Astro.
Given two identical frameworks, my preference would be for one that updated regularly but less frequently. Quarterly or even twice-yearly updates would be perfect (patches excluded). Enough to keep you thinking about updating dependencies but not so often that you’re always behind.
Sadly for me, that isn’t the reality with Astro. Is that enough to drive me to choose a different framework? No. But release velocity was not something I considered when choosing Astro. Now with a little more experience it is something that I will think about in the future.