Modernizing jDeepL: Java 11 to 17 and Automated Releases

2025-08-02T00:00:00+02:00

I had some time last night - the neighbors were having a party and I couldn't sleep - so I decided to tackle a task that's been sitting in my backlog for way too long: modernizing jDeepL, my unofficial DeepL translation app for Linux.

For those who don't know jDeepL, it's a simple JavaFX desktop application that lets you translate text using the DeepL API. Nothing fancy, just a basic GUI with source and target text areas, language selection, and the CTRL+C+C keyboard shortcut for quick translations. I built it because I wanted something native on Linux instead of using the web interface all the time.

The main problem was that the project was stuck on older dependencies and had practically non-existent documentation. This became obvious when people opened and commented GitHub issue #1 back in March 2023. Around that time jDeepL was also mentioned in a Golem.de Article and I received a couple of Emails. All pointing out that there wasn't much instruction on how to build it and that they ran into compatibility issues with newer Java versions. The project was using Java 11, JavaFX 11, and a bunch of outdated libraries that caused build failures for anyone trying to compile from source.

So last night I finally decided to modernize the whole thing a bit. The work took me a few hours of checking new versions, reading changelogs, and validating that updates wouldn't break anything. I bumped Java from 11 to 17 and updated all dependencies to their latest stable releases. So, same libraries I was already using, but brought up to current versions.

The most interesting challenge was fixing the FXTrayIcon dependency. It had a deprecated constructor that I was using, so I had to dig into the library's source code to understand what the deprecated builder actually did and how to achieve the same result with non-deprecated methods. The old approach let you define image scaling directly in the constructor. The new way requires a two-step approach: use the builder pattern, then call the image scaling method on the returned object. Small change, but it took some detective work to figure out the right replacement.

The documentation also needed some love. The README was more or less just a description of what the app does, with not much information about how to actually build or run it. I added a Quick Start section and expanded the Build section with Maven setup instructions. More importantly, I set up automated releases using GitHub Actions.

This is actually a huge step forward compared to how I handled releases before. Back in version 0.7.4, I was manually uploading pre-built JARs to my personal Nextcloud instance. This worked fine and was documented in the README, but wasn't exactly convenient. Now when I push a tag, the GitHub Actions workflow automatically builds the JAR and creates a GitHub release with the file attached. Users can just go to the GitHub releases page and download binaries instead of having to build from source.

One thing I discovered during this modernization is that jnativehook, the library I use for the global CTRL+C+C hotkey, is likely on its way to becoming unmaintained. It only works with X11 sessions, and with bleeding-edge distros pushing toward Wayland, I might need to find a different approach for hotkey handling in a few years. But that's a problem for future me.

The project now builds with Java 17, uses updated dependencies, has expanded documentation, and automated releases. The original issue about build problems should be resolved now. Users can either download pre-built JARs from GitHub releases or follow the updated build instructions to compile from the source code.

It wasn't groundbreaking work, overall just maintenance that was overdue. But sometimes that's what projects need, a bit of attention to keep them current and usable.

Next, I'm thinking about creating OS-specific packages using jpackage. This would let me provide .deb and .rpm packages that ship with all dependencies. This would make the usage even easier compared to the current JAR file approach.

jDeepL version 0.7.4 released, pre-build binaries available

2023-11-01T00:00:00+01:00

It's been a while since I announced and released the first version of jDeepL - Unofficial DeepL App for Linux based on JavaFX. Since then it was mentioned on a couple of news sites and people reached out to me. Mostly with pointing out that there are not much instructions how to build it. And that I don't offer pre-build binaries that people can directly run.

That was good and valid feedback. I was focused on sharing it with the world. I didn't take into account that non-tech people are the actual users.

As part of the recent Git repos moved to Sourcehut, with mirrors on Codeberg, GitLab and GitHub activities, I allocated a bit of time to tackle some of the addressed issues. The README should now be a bit clearer. I updated some dependencies, improved the release scripts and published v0.7.4.

For the future I plan a CI integration that takes care about publishing new versions. Until then, I uploaded the pre-build version to my personal Nextcloud instance: https://hub.xlii.cc/s/an7E9E9bgFBpRG3

Latest version: v0.7.4 (SHA256 checksum: 206c13ec820451096bdf07f29e336c6827c750db11554d9dcc68812c8b56c796)

It's nice to see that some people find useful what I created.

jDeepL - Unofficial DeepL App for Linux based on JavaFX

2022-05-30T00:00:00+02:00

I'm a daily DeepL User, the Windows App is convenient but nothing comparable was available for Linux, so I build my own.

jDeepL is a very simple GUI Application based on JavaFX, developed and tested on openSUSE Tumbleweed with KDE Plasma.

You can find the GPL v3 licensed Source Code here: https://dominik.wombacher.cc/~git/jDeepL/

It's in an early stage but covers the most important functionality: Translating the marked text by pressing CTRL + C + C.

At least that's for me the most important one from the official Windows App ;)

I will probably improve it further within the next time, feel free to Contribute.

Further details can be found in the README: https://dominik.wombacher.cc/~git/jDeepL/

The Wombelix Post - DeepL

Modernizing jDeepL: Java 11 to 17 and Automated Releases

jDeepL version 0.7.4 released, pre-build binaries available

jDeepL - Unofficial DeepL App for Linux based on JavaFX