Endive: A JVM native WebAssembly runtime | Hacker News Reader

Endive: A JVM native WebAssembly runtime

github.com

Shameless plug: we solved the opposite problem, running any Java application in the browser via WebAssembly: https://labs.leaningtech.com/blog/cheerpj-4.3

And yes, it does run Minecraft as well :-) https://browsercraft.cheerpj.com/

But can it run Endive?

I wonder what the performance loss is for each level?

Not something I tried myself, but conceptually if it's Java it should work.

CheerpJ Just-In-Time compiles Java bytecode at runtime, so it makes no difference if the classes come from JAR files or are dynamically generated.

How close are we to the birth and death of yavascript?

https://www.destroyallsoftware.com/talks/the-birth-and-death...

Mmhh should work for clojure as well then. I should try this.

I can confirm non-Java languages work as expected. I've personally tried Kotlin, and IIRC some user from the community reported Clojure to be working as well.

Consider joining our Discord for help: https://discord.leaningtech.com

This is a fork of Chicory, a bit more context of the relationship between the projects can be found here:

https://github.com/dylibso/chicory/issues/1296

Any background / context around what the Chicory author means in this comment?

> We'll consider merging in changes that make sense from Endive, but under the stewardship of the [Byte Code Alliance] I have very little faith in its future. My words mean nothing though having all but completely lost interest and use for WebAssembly.

What's the background / history of Byte Code Alliance?

Not everyone is on board with CORBA but with WebAssembly, which is basically what the reboot of the component model is, so probably that.

The AssemblyScript folks have a similar opinion.

yeah that is roughly the concern, many runtimes didn't want to continue on past wasip1, mostly because of the component model.

The comment is from the Dylibso CEO.

On the CNCF wasmCloud Community call this week we played with this: - a demonstration of Endive - implemented CNCF wasmCloud host - Integrated into Vert.x as an example

And discussed the roadmap.

Blogpost and video here: https://blog.cosmonic.com/engineering/2026-05-26-diving-into...

Lots of context for this project on the Bytecode Alliance blog: https://bytecodealliance.org/articles/endive-and-the-next-ch...

Ah, looks like they will be packaging a Rust runtime on top, not as interesting as I thought.

the "redline" compiler is cranelift which is written in rust, but i think it's being compiled to Wasm first then JVM bytecode so it still works zero dependency. not sure if that will continue to be the case.

Projects like this would be significantly funner and easier to make in Jdk25+(well technically 24+) because of the new Java classfile/bytecode API. It looks like Endive uses OW2 ASM, probably because this supports back to Jdk11. The new jdk API has a minimum target of Jdk17. OW2 ASM is significantly harder to use IMHO though.

What got me into this is I just finished a major release of Petrify (https://github.com/exabrial/petrify) that compiles ML Models to JVM Bytecode. It requires Jdk25 to do the compilation, but the compiled models can run on Jdk17+.

I'm looking for more side projects to use the classfile API on.

You might want to take a look at https://www.graalvm.org/webassembly. Runs on JDK25 and has very good guest<->host interop.

There is a comparison between the two here:

https://www.linkedin.com/feed/update/urn:li:activity:7465778...

bytecode manipulation has been a thing ever since late 90s and early 2000s (e.g. BCEL, Jan, 2001), along with byte code decompilation.

Generally one must understand how bytecode signatures, all flavors of invoke, and constant pool work. After that using visor pattern or 'functional' alike stuff makes no difference whatsoever.

I have used (still using) bytecode manipulation along with custom classloaders as part of my job (albeit not on daily basis any longer). Personally, I don't consider objectweb asm hard to use in any way. and java's class file won't be funnier - perhaps it was the very project I'd not pick bcel, though.

if i understand correctly, the new redline compiler doesn't have these limitations. it's not based on asm. (edit: but this hasn't been merged to mainline yet)

It will be really great if this becomes a second popular runtime with both GC and WASI component model support. Wasmtime being the only runtime with that combo is a bit concerning. Node supporting the component model will help a lot too.

The component model is still in phase 1 (standardization is phase 5) and the Bytecode Alliance are its sponsors and the ones pushing it into the ecosystem with wasmtime.

I don't think you're fully saying what you want to here. Are you saying this is bad?

The point of a component model is interoperability, so the more runtimes that support it the better.

I think just pointing out that it's still in stage 1 so it makes sense that it's not supported in every runtime yet

I am saying as far as anyone other that the Bytecode Alliance is concerned it is custom API for Bytecode Alliance projects.

Component model can run in any js runtime with jco

I guess we can come full circle and eventualy port it to Android Java.

Android is supported!

I was trying to write something like this in rust at some point, just for the joke that you can compile that rust to wasm, and then it can compile itself to JVM assembly. The complexity of it turned out to be quite a bit too high for a joke only.

Is this being handed over to the Bytecode Alliance or is this a hard fork and will diverge from Chicory? It isn't clear from the announcement but I suspect the former.

Yeah, this was the first thing that came to mind, how does this compare to the Truffle WASM implementation. The Graal Polyglot API is pretty incredible, we've been using it for a JavaScript/Python plugin system in a JVM app, and it's been amazing.

Agree about Graal being really good. there are some different use cases for embedding wasm in an application. chicory / endive is not just for embedding another language. the main use case was always secure plugin-systems. but there are other use cases. also it's not controlled by Oracle and works well outside of their ecosystem, which i think some people value. this question came up a few times and were addressed in talks and blog articles:

https://github.com/dylibso/chicory#on-the-press

If you haven't seen The Birth & Death of JavaScript, it's well worth a watch:

https://www.destroyallsoftware.com/talks/the-birth-and-death...

Why not resurrect applets? We had this webasm thing 30 years ago.

The two main points are that wasm is entirely sandboxed and that it's designed to be streamed, and to start up very quickly. The official Java youtube channel coincidentally posted this two days ago - https://www.youtube.com/watch?v=fy0KyGLrbJo - which includes some interesting details.

Another Shameless plug: A common interface for webassembly engines, including Chicory, in Java https://github.com/tegmentum/webassembly4j

I wrote this for a number of reasons but oddly enough Java probably has the most options as far as runtimes go but the support is pretty fractured. The underlying runtime adapters add support for wasmtime and wamr as well. There are two JNI implementations for wasmtime but they are badly out of date and haven't been updated in years. There are a number of reasons you might want to use one runtime over another and this was supposed to unify them under a single interface and make swapping them out easy. It also allows you to easily compare the tradeoffs of various engines.

Finally we can run Kotlin/WASM on desktop! /s

Reply on news.ycombinator.com