logoalt Hacker News

yogthos • yesterday at 10:49 PM • 0 replies • view on HN

Having a portable Clojure API that's decoupled from the JVM would be ideal going forward. That would make life much simpler for both dialect and library maintainers.

In my opinion, the reader conditional approach simply doesn't scale because it puts the burden on the library author to maintain conditionals for each dialect and to understand how the dialects work internally. It's clearly a better approach to have a common API that libraries target, and then each dialect can implement in the way that makes sense for it.

I suspect we're still a little while away from that though, and in the meantime creating Java shims seems to be a reasonable approach to take in order to support existing code. That can also help inform what the Clojure centric API needs to cover which is an additional benefit of doing this work.

While I agree with you in a philosophical sense that Clojure as a language should not be viewed solely through the lens of its existing host platform, there is the practical aspect of being able to leverage the existing ecosystem from a particular dialect. I'm using Jolt for my own work and personal projects, so I want to be able to use mature libraries that I'm familiar with like Ring, Reitit, Malli, Selmer, and so on. These allow me to build real world projects without having to bootstrap the whole ecosystem from ground up.

And to the final point, I've always been up front about LLM usage in the project, and Jolt now has a clear AI policy in the repo as well that people can read: https://github.com/jolt-lang/jolt/blob/main/AI_POLICY.md

So, I would hope that nobody feels deceived in any way here.