babashka 2026-11-14

hi, is it planned that babashka drops the jvm requirement to resolve dependencies? I see that there are some cases where the jvm resolver is still required. I’m asking because I co-maintain the babashka nix package, and currently we have a hard dependency on the jdk, just wondering if at some point in the future we can remove the jvm dependency.

@jlle This should already be the case, at least for bb’s own scripts. Java is still necessary of course when you launch clojure itself using babashka.

I thought s3 were not supported

true. at least not today. is anyone using that?

I can do a poll. I haven’t had anyone ask about this

Not sure how many people use it, but if that is still an option, I think we should maintain the dependency in nix. We also offer a version without the jvm , but it isn't the default one

I can see how much work it would be (or how much binay size it would add) to add s3 support

can people use bb on nix and get java in addition optionally? e.g. if the want to launch clojure

Not really needed, currently you have both options on nix, with or without the jdk dependency. I was wondering if it would be a good idea to remove the jdk dependency option, I think it is okay to keep it as it is, I don't want to give you extra work for an edge case. I would think about switching the default, so no hard jdk dependency on the jdk unless you opt it.

that makes sense to me

Poll: is anyone using S3 repos for their babashka or clojure deps? See https://clojurians.slack.com/archives/CLX41ASCS/p1794673729671859 👍 👎

👎 7
👎🏻 1
Alex Miller (Clojure team) 2026-11-14T17:43:59.608079Z

I had the same question a couple months ago, asked in #C6QH853H8 I think. I did not find anyone using s3 deps other than Daromic users (and I think that stuff is all in central or clojars since it went free)

Alex Miller (Clojure team) 2026-11-14T17:45:25.461429Z

I broke all the s3 stuff out into a separate dep and it only installs if you have the aws libs in the classpath, which the CLI does, but optional for lib use

we do at metabase but we do all of the steps to ensure that it appears as a maven repo. But a raw s3 repo would be super nice:

"metabase-maven-downloads" {:url ""}

in particular i think we redistribute amazon’s athena jdbc driver

> we do at metabase but we do all of the steps to ensure that it appears as a maven repo. do you do this because bb’s non-JVM resolver doesn’t have s3 support or did you have other reasons?

Just wanted the easiest way to get jars to the clojure cli. I forget details but my memory is that we had to follow the metadata layout but otherwise it's just an s3 bucket with a particular shape?

that works as a public mvn repo yes