datalevin

Jon Hancock 2025-12-09T06:36:55.530079Z

Any thoughts on if datalevin would work with jank?

👀 2
Jon Hancock 2025-12-27T08:43:29.568589Z

thanks @huahaiy. I enjoy working with datalevin. I'm not in a rush to leave the jvm.

🙏 1
christos 2025-12-17T08:04:35.484959Z

maybe wait of jai to be released

Huahai 2025-12-10T15:54:54.720119Z

Not right now, as we use a lot of Java code. However, Jank is a potential target when we decide to move more code to native. We will stay on JVM Clojure to take advantage of the mature environment for exploratory programming. We will move to native when all design choices are worked out.

👍 1
Huahai 2025-12-10T17:13:23.385809Z

Basically, features are first explored and put to use in JVM Clojure. After a few rounds of real word use and hardening, we settle down an implementation and push it down to native when appropriate. E.g. all our index iterators, samplers and counters were first written in Clojure, now they are written in C. The push down is surprisingly easy.

Huahai 2025-12-11T01:25:51.563739Z

My plan is to eventually push most functionalities down to native code, while keeping the same Java/Clojure API. That makes it easier to provide language bindings for native languages, e.g. python, typescript, etc. Jank could be our path towards that, depending on what it eventually becomes. Zig, Odin, and C3 are the other options I am looking at.