Regular update, May 2025

[konsoletyper]

Hi. I think I do a lot of purely technical work, which shows good result - after two releases and less than a year of work we finally have new WebAssembly GC backend that just works. However, I always a bit reluctant to do community work. I'd like to find a way to change it. So I'll try this new format: on some regular basis (I'm not sure, perhaps it's monthly, or two times a month, or may be just "when I have enough time and motivation", but not too rare) I'll share my progress, my thoughts and my ideas and my plans here. Let's start.

First of all, my current focus is to create an impressive demo. Current ones that I have are two simple and too small, so as anyone sees them, they think that TeaVM is a toy project. Another reason is that currently I'm too bound to my employer's project when I test new releases; I want to have my own "big project". So the decision was to revive my old project of translating Java compiler to WebAssembly. Many years ago there used to be a page on TeaVM's site where everyone could just type any JavaScript code, press "play" button and get the result. Unlike most of such playgrounds, my version did not do any roundtrip to the sever for compilation - everything was done by the client. However, after Java 8 there were too many show stoppers. First of all, modularization. Second, javac started to use java.nio.file package, which is huge and which is not implemented in Harmony project, so I have to re-implement it myself. Now I did just that: implemented most of java.nio.file, and now I can at least compile javac 21 (but it still does not run). I'll keep sharing my progress on this project. Another possible benefit from such playground is documentation: I'm going to prepare code samples for playground that cover specific topics (say, JavaScript interop) and set permanent links for them, so that I could put these links into documentation.

Second, by release 0.13.0 I'm going to close final gap between JS and WebAssembly GC backend, namely, support for green threads. In addition, if this won't take too much time, I hope to extend reflection support: add support for annotation on fields and methods and perhaps even implement generics in reflection and support proxies.

Finally, I have always realized that keeping this state when TeaVM is my personal hobby project is a way nowhere. Sometimes I make efforts to change this - and always fail. But I'm going to keep trying. My last effort was 1.5 years ago when I started to thank PR authors in release notes and encouraged to make contributions. I also added corresponding section on the web side. This however did not bring good experience. Partially, this is my fault - I'm usually reluctant to review PRs. Partially this is because I felt that I need WebAssembly GC backend, because it seemed a year ago that it's the direction where the world is going (and I think that decision was right). And this made me sink deeply into coding, so I became even more unresponsive. I'd like to change different approaches of growing long-term co-contributors for this project, I hope to publish details a bit later.

Also, one small thing. Recently I added to examples to TeaVM's gallery sections:

• Commodore 64 emulator by @kenchis
• Masamune by @Quillraven

One special thing about them that they weren't created by me. And what I'm proud to find out: when I released TeaVM whith stable WebAssembly GC support, @kenchis just took his project and re-compiled it to WebAssembly GC without any assistance from my side. So I updated the link to point to WebAssembly GC version.

Let this format won't be monologue. Just communicate with me and with each other. Share your ideas, your progress, your projects. I'd like to encourage you to post here, in "Discussions" and in discord. I receive questions here - good, but don't limit your posts to questions. Share everything you want to share. And of course, if you have something to reply here - just write your replies to this post.

[Ihromant]

If you lack of demos, I have Heroes of might and magic III combat emulator here https://ihromant.github.io/ Unfortunately, I don't want to make this code public at this moment of time, so you have to trust (and you have seen it) that it's implemented using TeaVM.
I don't have anything to blame, for my needs TeaVM fits perfectly and every small snippets I need, I'm implementing using it. So next ideas are just like "it would be nice to have".

1. When I did small snippet with charts, I used Google Charts library. And I understood that it would be great to have TeaVM wrappers for popular JS libraries. Still, anyone can create TeaVM wrappers for them and share, using JS interop it is simple.
2. This is just crazy one and I assume that it's too much effort, but you have C backend. If create wrappers for Objective-C native classes and have Objective-C backend, you will be able to implement business logic for mobile devices in Java provided you are familiar with native API.
   Thanks for your efforts and nicely done.

[konsoletyper]

If you lack of demos, I have Heroes of might and magic III combat emulator here https://ihromant.github.io/

Impressive! Yes, I knew that you are doing something like this, but then you were on the very early stages. Good to know that you keep working on the project. BTW, why did you remove global strategic view?

Unfortunately, I don't want to make this code public at this moment of time

Why? Are you going to make a business out of your project ultimately?

When I did small snippet with charts, I used Google Charts library. And I understood that it would be great to have TeaVM wrappers for popular JS libraries. Still, anyone can create TeaVM wrappers for them and share, using JS interop it is simple.

Agree. But usually I just don't have enough time and motivation to write bindings. However, if someone wants to champion this, you are welcome. But please, it should not be in the format "I send you unexpected PRs". It could be collaboration, with discussion in private chats, even calls.

This is just crazy one and I assume that it's too much effort, but you have C backend. If create wrappers for Objective-C native classes and have Objective-C backend, you will be able to implement business logic for mobile devices in Java provided you are familiar with native API.

There's one problem with this: I don't have any Apple equipment at home. Buying an expensive laptop and an expensive phone just to from time to time contribute to my small hobby project would be a strange idea. Especially given that C backend, and specifically iOS C backend are of no priority to me. Also, one subjective factor: I'm used to PC keyboard and PC hotkeys (specifically, KDE) and to KDE UX, so every time I try to sit after macbook, I realize that I 10x less productive than with my usual environment.

But if you want to champion this, we can discuss this and you can start leading this direction. This, however, includes not only "create a prototype", but to "write documentation", "write examples", "post to blogs", "support users". Also, this also should not be unexpected PRs, bur rather collaboration with discussions and join decision making

[Ihromant]

Impressive! Yes, I knew that you are doing something like this, but then you were on the very early stages. Good to know that you keep working on the project. BTW, why did you remove global strategic view?

Actually everything is opposite. ihromant.github.io is actually a prototype (there is no backend and everything is in-memory) which I maintained until I started implementing backend together with global strategic view. If you check commits in master, you can observe that last meaningful commit in master there was in 2023.

Why? Are you going to make a business out of your project ultimately?

There are enough open-source projects which are related to Heroes III (for instance, WoG, VCMI). Also there is funny phenomenon: approximately once per 2 years some russian guy (for some reason this game is popular in post-soviet area) posts his raw efforts on Habr and asks for community support. And after that he stops any activity on that project. Last that case as I remember had name "herowo". And just after posting it on habr author stopped any activity on that project :D (still, you can find it in google and even "play"). So, the reason is simple: to maintain discipline understanding that nobody other than me can finish this project.

There's one problem with this: I don't have any Apple equipment at home. Buying an expensive laptop and an expensive phone just to from time to time contribute to my small hobby project would be a strange idea. Especially given that C backend, and specifically iOS C backend are of no priority to me. Also, one subjective factor: I'm used to PC keyboard and PC hotkeys (specifically, KDE) and to KDE UX, so every time I try to sit after macbook, I realize that I 10x less productive than with my usual environment.

But if you want to champion this, we can discuss this and you can start leading this direction. This, however, includes not only "create a prototype", but to "write documentation", "write examples", "post to blogs", "support users". Also, this also should not be unexpected PRs, bur rather collaboration with discussions and join decision making

Same situation for me. I worked for 1.5 year on Macbook laptop and I don't understand UX fans of Apple. For me this time was counterproductive and I was just forced to use Macbook for my work despite I asked several times "anyway we have Linux environment on servers, why can't I use any commercial "certified" Linux distribution?".

So, there were just few ideas to share (like usually it's done in "forums" :) ), currently I'm trying to make MVP from this simulator and restored my mathematical studies that I dropped after defending PhD thesis. These activities currently consume all my free time, so I don't think that I'll be able to participate actively in open-source project in closest future.

[kerbymart]

Hi, discord invite like is expired.

[konsoletyper]

Hi. Thanks for the info. Issued a new link

[Ihromant]

Other idea came to my mind. But I assume that implementing it would be not worth spending efforts or impossible, therefore sharing it here first.
There are some functions in Array prototype that correspond to utility methods in Arrays. For instance:
https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Array/sort <-> Arrays.sort
https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Array/fill <-> Arrays.fill
https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Array/slice <-> Arrays.copyOfRange
But I understand that arrays in TeaVM are represented not as simple arrays, so my question is whether it is possible and worth implementing?

[konsoletyper]

It's possible and worth implementing, but only in JS backend, since in Wasm any call between JS and Wasm is expensive. Also, please note that there are other backends that can't rely on JS API. Finally, primitive arrays are not Array.