Zemer Bugs Comments and Feedback

And as far as YouTube API RE, I take browser cookies from a YouTube account, and use my terminal to hit YouTube with everything I’ve got, from every angle, for HOURS, DAYS, I don’t know anymore… Until I get what I want.

Who do you think does most of the legwork when I document exactly everything I need? Claude.

How do you think I documented the exact way, which headers, which fields, which tokens etc… to successfully query YouTube? Trying EVERY way, with AI assistance, until I had a bulletproof API query mechanism.

5 Likes

What do you think I told one of the developers of metrolist who was having an issue with part of the API?

Here’s a Claude guest pass, use cookies, use your existing knowledge, and here’s a doc or two to start you off.

Shoin.

Claude+Existing knowledge, from where? Claude testing.

Brain go brr with new knowledge.

Repeat.

2 Likes

Yes, I still sometimes scratch my head when I realize I made a stupid mistake. I move on and learn from it.

I look at my codebases from a year ago and laugh. That will probably happen next year too, but the laugh won’t be as loud.

Rant over. Exposed :slight_smile:

Don’t care.

5 Likes

That was great and informative

Off topic but since you mentioned it, firstly I’m not a spender, secondly, monthly subscriptions is scary…

I do very similar. That’s how I know the little bit that I do. I use AI more and more in such a manner but not like
Fix it
Fix it
Fix it
But as you wrote, understand and learn and have AI as an assistant. Don’t let it replace you because it can’t (yet)

I’m waiting this because I think many users here already use AI. The issue isn’t that they should start using it but rather that they should stop trying to have it replace themselves.

But I think this can be moved to other help and opinions about AI and learning topics. Not really my point that you’re ahead of me (although it’s definitely a part of it and doesn’t change with this).

My main point was that you already put in TONS of time into it. For an outsider to come in, learn this specific codebase, what every piece of code does, why it’s necessary, why you chose one way over the other etc etc is a huge task and even once he thinks he gets it and makes a pr, since he’s not you and has his own views, he may fix stuff in a way you don’t like, like making a architectural decision you don’t agree with.

Onboarding others to your project isn’t that simple is my point. When there’s some issue, just saying can someone please try to fix it isn’t always actually doable. That’s my point.

The solution? I don’t know. But perhaps distributing the work and getting different people involved into different parts of the codebase. Like explaining to the people you’re onboarding exactly your vision, how the code does what it does, why you chose whatever you did etc till they can easily take over that part (obviously you can still supervise and review before merging…)

I’m not sure if this is the correct solution but that’s my point.. hope it’s clearer now.

(On a much smaller scale, someone gave me code to enhance my email to sms bot. He used AI. There were great parts to it but it changed the structure too much and does much too many stuff that I don’t like at all. I felt that if I wanna implement those ideas I’ll probably rewrite all that myself even though it’ll take much longer. That’s a classic example of others not understanding your codebase)

TLDR

Your codebase is huge and complex with much experience, decisions etc which make it

  1. Hard to come in understand and fix
  2. Often useless as it may not align with what you want cuz the outsider doesn’t know your reasoning behind your decisions…

100 percent

You are correct and incorrect. You do not need to understand the whole codebase to fix a UI issue, for example.

If people didn’t contribute to open source projects because of this, nothing would ever get done.

I am open to this. Alot of documentation is missing in Zemer because it started when I didn’t know as much as I do now. I’m fully open to creating a detailed wiki.


Now, it’s time to go shluffy.

1 Like

Funnily enough, I’m not either. I find it hard to buy myself things. I have never regretted (or been net red from) paying for my subscription. It has only advanced me to new opportunities.

1 Like

True but I hate UI and I’m more into backend and architecture and reverse engineering so at least for myself… Yes I guess for small UI style stuff it can usually be done. (Although major UI improvements will still have the same issue.)

I hear you. But firstly most open source projects that get contributions are much more documented, secondly they usually get contributions from people who took the time and effort to fully learn whatever part of the codebase they’re working on thoroughly. Which is why there aren’t hundreds and hundreds of contributors per foss big project. The amount of people that care enough and are savvy enough for zemer is a bit small…

Enjoy! Nice chatting :rofl:
To be continued tomorrow after 8pm (unless it’s quiet at work today which it won’t be)

1 Like

Last thing :rofl:

Get to work. I even gave you the source haha

1 Like

Waiting for

I’m also not sure exactly which parts can get improved but that we’ll figure out.

2 Likes

Touche

The problem with accurately documenting YouTube API is that it changes every 5 minutes, but I can do basics.

1 Like

:+1:

Agreed

Oops

Some people pay a few thousand per month for Ai

So why did you close the pr for the UI fix on Zemer?

Because you didn’t think at all when making changes, and you surely did not test it. Hopes and dreams that it will work.

Yes I did, It was a 1 line change adding a scroll option for small screens

And how do you know it works?

Because Claude Opus said that it would :blush:

1 Like

And that is why I closed your PR.

3 Likes

Someone asked on mitmachim if it was possible to add Skazka’s channel

And I asked on Jtech if its possible to add a different one.

They are not adding artists for now

1 Like