Live data from Hacker News

Tell HN: 2 years building a kids audio app as a solo dev – lessons learned

news.ycombinator.com

61–70 of 84 posts

Re: Tell HN: 2 years building a kids audio app as a solo dev – lessons learned

#61

Similar story here , but started mine when first Android phones were released. Had great success. And still have. Now with AI I have 2 max accounts with Claude and I don't touch any code anymore. I went full high risk cowboy style. All code. Server management, databases, security, upgrades, root access. Access to all my accounts, keys, hashes all goes into my prompts. Everything with ai. I don't even go to the Playst…

That's bold! Full AI-driven, including deploys and server access – respect for going all in. Curious how you handle when AI makes mistakes on production systems?

I think I now went all in about 2, 3 months ago. It does make a lot of mistakes and its cowboy style 4.0. I cant remember a case were everything became a broken chaos. I think the AI does a much better job in maintaining my machines than i could do. Its able to configure low level stuff like fail2ban, iftables perfectly fine. Its much better in reading logs or solving issues. Once in a while it makes a mistake, like misconfiguring firewal rules and lock me out. All solvable issues. Software wise its also better. Yes you are prompting in circles sometimes. But thats fine. Usually it will fix the issue I have after a few iterations. Its not so much different that normal development, but on a higher level. You are dealing with a senior developer behaving like a child.

Re: Tell HN: 2 years building a kids audio app as a solo dev – lessons learned

#62

> SwiftUI is a joy to work with and performance is great. It really is, unless you want to draw outside the lines . If I want to do anything that isn’t explicitly in the SwiftUI model, it becomes Kludgeopolis, very quickly. There’s some things that can only be done with UIViewRepresentable, which I consider to be a bit of a “white flag.” Also, because of the way that the rules are written, it’s difficult for me to se…

Totally agree – it can get messy once you go off the beaten path. But for most things you can build so much with just SwiftUI and a few Apple frameworks. No node_modules with 500+ dependencies. Coming from web dev that still feels like a luxury.

Re: Tell HN: 2 years building a kids audio app as a solo dev – lessons learned

#63
This looks very nice, I love that you're making a real solution to avoid the dark patterns of companies like Spotify. I used Subsonic for a while as a simple way to curate music for kids and keep them off Spotify, but eventually my oldest got past the "listen to the same album over and over again" stage, and adding everything to Subsonic became a chore, so we succumbed to giving her an account. Yet I really don't like how Spotify has been adding direct messaging, Wrapped, Jams, discovery etc. – stuff that on the face of it looks fun (for adults), but when you see the addictive effect it has on the tweens it's just horrible. Technically I guess Spotify is now illegal for For my youngest, I installed Qobuz, which has all the music, but much fewer features. They reportedly pay artists more per stream than any of the other apps (I guess because they don't have so many expenses for social media app development). Your app UI looks much cleaner though :)

Re: Tell HN: 2 years building a kids audio app as a solo dev – lessons learned

#64

This looks very nice, I love that you're making a real solution to avoid the dark patterns of companies like Spotify. I used Subsonic for a while as a simple way to curate music for kids and keep them off Spotify, but eventually my oldest got past the "listen to the same album over and over again" stage, and adding everything to Subsonic became a chore, so we succumbed to giving her an account. Yet I really don't lik…

Thanks! Yeah the dark patterns in streaming apps are rough, especially for kids. Wrapped and all that social stuff is designed to keep you hooked.

Give Muky a try if you want! Qobuz is nice – would love to add more services like Qobuz or Deezer at some point, as long as the API is good and offline listening is supported. And yeah, wish more services would adopt a user-centric payment model – would be so much better for artists.

Re: Tell HN: 2 years building a kids audio app as a solo dev – lessons learned

#65
YouTube music support? (ᵕ•_•)

I set a (very) old phone into child mode total lockdown with only YouTube music installed to make use of my family account and streaming playlists ... only to find it (the YTM app) wouldn't open due to the social media ban in Australia (legislation which I otherwise - controversially - completely support).

Re: Tell HN: 2 years building a kids audio app as a solo dev – lessons learned

#66

> SwiftUI is a joy to work with and performance is great. It really is, unless you want to draw outside the lines . If I want to do anything that isn’t explicitly in the SwiftUI model, it becomes Kludgeopolis, very quickly. There’s some things that can only be done with UIViewRepresentable, which I consider to be a bit of a “white flag.” Also, because of the way that the rules are written, it’s difficult for me to se…

Totally agree – it can get messy once you go off the beaten path. But for most things you can build so much with just SwiftUI and a few Apple frameworks. No node_modules with 500+ dependencies. Coming from web dev that still feels like a luxury.

It’s easy to build high performance custom components, for example with Canvas which is excellent. However, at the top level of an app it’s got to be the standard way.

That may be a good thing for usability across apps but it feels like a low code platform sometimes.

Re: Tell HN: 2 years building a kids audio app as a solo dev – lessons learned

#68
post #67

> ... if Apple grants me the entitlement. I have a sneaking suspicion they might steal your idea and release their own version before they do that. But maybe you get lucky and they don't do that, so definitely go for it. Best of luck!

Honestly, I don't think Apple would bother with this niche. They have biggAIr priorities. And if they ever did, it would validate the idea - and they'd probably do it differently anyway.

For now I'll keep building. Worst case, my kids still have a music player that works exactly how we want it. :)

Re: Tell HN: 2 years building a kids audio app as a solo dev – lessons learned

#69
post #65

YouTube music support? (ᵕ•_•) I set a (very) old phone into child mode total lockdown with only YouTube music installed to make use of my family account and streaming playlists ... only to find it (the YTM app) wouldn't open due to the social media ban in Australia (legislation which I otherwise - controversially - completely support).

YouTube Music would be great, but there's no official API for third-party apps. Only unofficial libraries that scrape the web interface – not something I'd want to build an app on. Too fragile and could break anytime.

Interesting about Australia – sounds like the social media ban for under-16s now includes YouTube, but YouTube Music as a pure streaming service should be separate. The ban targets social platforms with feeds and interactions. Might be worth checking if YouTube Music specifically falls under it or if the device lockdown is catching it by accident.

Re: Tell HN: 2 years building a kids audio app as a solo dev – lessons learned

#70

Earlier quoted context omitted.

Totally agree – it can get messy once you go off the beaten path. But for most things you can build so much with just SwiftUI and a few Apple frameworks. No node_modules with 500+ dependencies. Coming from web dev that still feels like a luxury.

It’s easy to build high performance custom components, for example with Canvas which is excellent. However, at the top level of an app it’s got to be the standard way. That may be a good thing for usability across apps but it feels like a low code platform sometimes.

Know what you mean. SwiftUI can feel similar – great for standard patterns, but you hit walls when you want something custom. Ended up mixing in UIKit for some edge cases.

Trade-off I'm willing to make though. For a solo project, fast iteration beats pixel-perfect control.

Post reply on HN