LISA code, vibe code, LISA List2 and GitHub??

[bmwcyclist]

***important note***
I am NOT a developer. I am not even a good scripter; I typically have to pull out a reference to write a BASH shell script, let alone touch anything that needs to be compiled. I think the last programming language I was fluent in was AppleSoft BASIC, and my actual developer friends tell me that is not even programming because I did not compile it...

I am a computer nerd and an engineer but not a developer.
****

So my question (s) for you guys that have actual experience:

I want some things to use in the LOS and if no one else is doing it... that leaves... ME. and my ME I mean me an Claude...

Some people embrace vibe code; some don't...

After Alex made LOS Minesweeper I got inspired and have been spending all of my lunch money on Claud tokens...

I only have one small thing to show for it so far (see yesterday's post), but I hope to do a few more.

Some people have told me I should make a GitHub account, but I know some friends who say vibe code shouldn't be there. Should I only publish here on LisaList2?

I am not good at keeping up with old projects, so I think most of what I make will be abandonware (assuming that term applies also to code put in the public domain).

Thoughts? Comments? This is all the wild wild west for me.




.

[stepleton]

There are a lot of people with a lot of opinions about vibe coding. I won't prescribe any particular point of view myself, but here are some facts that might be relevant:

First, github is full of partially and fully vibe-coded projects, and github for its part seems happy to encourage this. I see a "Try GitHub Copilot" button on github's frontpage. Alternative hosts like codeberg have a rule against vibe-coded projects, but those hosts are not github.

my actual developer friends tell me that is not even programming because I did not compile it...

OK, I was wrong, I will prescribe at least one point of view. Your "developer" friends are silly to make this distinction. There is no clear line to draw between programming and not-programming and peoples' opinions will vary on what counts, but "compilation" as the dividing line is not one that's entertained by anyone serious, nor (I suspect) is it one your friends could actually define with much precision. (Though part of me would love to hear their attempt.) I would not waste much time debating or even discussing this with them further.

Back to facts. Coding agents reduce the need to understand as much about the problem you're trying to solve and the methods being used to solve it. This allows for rapid progress but also sidesteps opportunities to learn and to make connections. In your case, you learned about relevant references and similar projects after the fact. This might be fine, but if nobody had spoken up, then you might still be in the dark about those things, they wouldn't have the chance to inspire or interest you (which is not to say that they did), and you wouldn't be able to pass on that knowledge to anyone else.

Doing your own study might give you the chance to make different choices than you would otherwise. For example, your tool works with the Lisa OS filesystem, but did you know there are at least three versions of the filesystem? The B-tree based catalogue is only used on version 3 of the filesystem: earlier versions had different underlying structures. Now that you know that your tool is only Workshop 3.0 compatible, what might you do differently? What might you have done differently if you had known this at the beginning, along with all the other background info we've discussed?

You can use coding agents as learning tools. For example, you may not know what a B-tree is or why someone might want to use this data structure as the basis for a file catalogue. (Plenty of people got advanced computer science degrees without ever having to learn this; I certainly never had it taught in my own schooling.) Claude will explain this to you with as much patience as you can afford. Repeated chats like this could unlock a lot of new knowledge for you and may ultimately help train a kind of "programmer's instinct" for thinking about programming in general --- and that will help you use coding agents better. But you can also get that same kind of knowledge by talking to programmers about programming (maybe look beyond your developer friends here though...) or by reading books, watching good youtube videos, etc.

So I will (oops!) venture a second opinion here: if you do choose to use coding agents, ask the agents to explain to you what they are doing. Ask follow-up questions. Ask them to speculate and explain how others might have done it differently, or even how an experienced person might have themselves prompted a coding agent to do the same job.

People have concerns about the social and environmental impacts of using coding agents or other large-scale AI systems. You may wish to research these concerns to determine whether you can use AI systems yourself in good conscience. You will find people with strong views on all sides of this debate.

One side remark: "Public domain" doesn't mean abandonware, it's just a statement about who owns something. Putting something into the public domain means relinquishing your ownership of that thing and declaring therefore that nobody owns it. You (or anyone) can still choose to maintain it, or not. SQLite is a public-domain software library that will be present in every modern computer or phone that you use and is very much maintained.