My music setup was a pinned browser tab for years. It worked, technically. It also ate RAM doing nothing, my media keys only worked when that tab happened to be focused, and I killed the music at least once a week by closing the wrong window. There’s no official YouTube Music desktop app on Linux, and most of what exists is the same web player in an Electron shell.
So I built my own. It’s called Limusic. Rust and Tauri, no bundled Chromium, audio goes through libmpv.
You sign in with your actual YouTube Music account, so your playlists, likes and subscriptions are just there, and liking a song or adding it to a playlist syncs back like you’d expect.
The part I actually care about is that it’s not bare bones:
- Proper MPRIS, so media keys and your now playing widget just work
- System tray, closing the window doesn’t kill the song, plus play/pause and skip from the tray
- Discord Rich Presence with album art and a live progress bar
- Last.fm scrobbling, connect once and forget about it
- Synced lyrics in a side panel, click a line to jump there
- Listen Together, synced rooms so you and a friend hear the same thing at the same time
- 8 themes, including Catppuccin, because obviously
- Gapless playback, volume normalization, shuffle, repeat, autoplay radio
- Your queue survives a restart
Free and open source, GPL. AppImage (self updating) or an rpm for Fedora, which is what I daily drive. No Flatpak or AUR yet, both are on the list. There’s a Windows build too, and no macOS one yet.
Would really love for people to try it, and I’d genuinely like to hear what everyone thinks. ❤️
Site: https://simohypers.github.io/limusic/ Downloads: https://github.com/SimoHypers/limusic/releases/latest
Not affiliated with Google or YouTube, just a thing I made because I wanted it
anecdotally, I’ve built a lot of projects and abandoned them because I ran out of time. With an LLM to reduce the input effort I actually think I would’ve maintained them longer.
do you have a source which backs up what you’re claiming?
Ok, so there’s no statistically rigorous review yet, but 73% of LLM-generated apps submitted to FlatHub were abandoned within 6 months. This is about the abandonment rate you should expect to see on any newly announced LLM-written project, and this figure appears to be much higher than hand-written projects (e.g. this study, while not directly comparable, shows a 16% abandonment rate).
The reason for this is sort of obvious to me: an LLM makes it really easy to write code that makes an app work, therefore the threshold of involvement/interest needed to “make an app” is much lower. This in turn removes the selection force that makes software authors likely to be interested in, and thus put in the effort (or tokens) to maintain their software going forward. (this is putting aside the fact that LLMs are still horrible at software architecture and quite bad at infrastructure, making the long-term maintenance more difficult).
I don’t doubt that some people will have a lot of passion for their vibecoded thing, and will keep it going for a while. This might even be the case here given the author went out of their way to post about it on Lemmy and respond to comments, seemingly without an LLM in the loop at that stage of the process. But it’s also important to disclose your LLM use so others can adjust their statistical expectations.
Likely you would have also built more projects if you had access to an LLM, thus giving you less time to spend maintaining each individual one. Which is kind of the problem here…
The bigger a project gets the more of an issue it becomes too
Fair points, I concede.
As someone who has built a lot of software it’s hard to see people treat LLMs as useless when they are shooting themselves in the foot by doing so.
Useless? No. Look at the Linux kernel, nothing is written by “AI” but adjacent llm tools are used. It depends entirely on what and how.
If something is vibe coded it can fuck all the way off. If they use a llm to point a human at possible issues (and the human knows what they’re doing) then the worst that happens is the human wastes some time chasing a nonexistent bug.
Ngl I was doing that for a while, google search is ass these days, even 4 years ago, ai was more useful for finding sources than google. I was using bingai at the time and I would also google and try to find sources the regular way, I had way more success just telling ai to only return edu sources (or similar instructions, idr exactly since it’s been a while) People that seek easy results with no effort could’ve got them through chegg back before ai really took off. It was always an active choice to learn how to do a problem rather than just grabbing an answer and faking the proof. (Ngl mostly because exams are proctored or in person, but that hasn’t changed even with AI, students will still fail to passs their classes and get degrees if they rely purely on AI)
Not the person you’re replying to, but
There are no shortages of evidence showing heavy LLM usage causing cognitive decline.
Likewise, if we accept anecdotal evidence, many users report losing understanding of their codebase after letting an LLM do substantial changes for them. Architecture drift is a common problem, too.
LLMs are trained on public code, much of it subpar, so it stands to reason that they are prone to regurgitating security vulnerabilities.
I am not including the plethora of ethical issues with LLMs, which I assume you do not care about coming from a purely utilitarian although short-sighted perspective.