I wonder if we shouldn’t support old (abandoned) projects using AI. This could be a separate F-Droid repository if someone is not inclined to use projects developed by AI.
For example, I took a game I’m interested in - Taps of Fire. I develop “free” songs for Frets on Fire and Frets on Fire X. It would be nice if the game were also available on Android.
So I decided to use Gemini 2.5 Pro with Gemini CLI to build the game on new libraries and so on. I’m not a Java programmer (etc.), so AI was supposed to do everything - move the game from Ant to Gradle, adapt it to the new Android requirements.
And it did. It took about 20 minutes and a few tests, but the game is now playable on new Android 15 hardware.
Maybe we (especially people like me) should improve games and programs using AI (especially games, in my opinion, because they don’t age) to run on new Androids, and at most, packagers should prepare applications to build and add them to the F-Droid repository.
For now, I’ll try to add the option to pause a song. Unfortunately, it crashes then.
I’ll write a post. Basically, I previously made a “fork” of the game Frets on Fire X (FoFiX). The original version works well with Python 3.7, I think. Together with AI, we managed to get it to version 3.13 (i.e., the current version).
Where did this idea come from? Generally, I have been trying to develop FOSS games for years. But not by creating them from scratch, but by taking game engines or old FOSS games and improving them.
Generally, many people involved with FOSS don’t like AI. Of course, I prefer open models, but unfortunately, closed AI is often better. So I decided to try to develop a few games further, using AI.
The results are generally positive, but what you have to remember is that AI can write code correctly, everything works, but suddenly it goes “poof” and all the code is useless.
Gemini CLI is a variant of the Gemini client that can run applications on the user’s computer. So, in addition to rewriting code, it can build it and, if it encounters an error, correct it, and so on and so forth.
I will write a whole blog post about it, because the topic seems interesting.
I would have to take a look, actually. I just downloaded the source from F-Droid (the project isn’t on GitHub, so I couldn’t fork it) and worked on that.
