A friend of mine asked for my advice about #codingforkids .
He's a teacher on #elementaryschool and wants to start a small #coding project.
I can remember software witch uses coding-blocks for childeren to build small games.
A quick search on #google gave me #codemonkey
Not sure if this is what I'm looking for but we'll see.
When the dinosaurs roamed the earth and I was a #codemonkey and had to earn a crust bit flipping...
... I have developed eleventh commandment. Purely because of hours(days/weeks) of wasted time, chasing bugs that were not there.
The 11th commandment is;
👉 CHECK YOUR DATA 1ST👈
And so, I have burned two full compute cycles, trying to render a skull with the interior 'illuminated' with no light (0.0) on the interior faces.
The problem: Back-face darkening only works for faces pointing AWAY from camera. But inside a hollow object, the far wall faces TOWARD you.
We need the original geometric approach - but it failed for concave meshes.
When in #Blender (a reminder I have nearly 0 working knowledge with it), I found that both the inner and outer faces are identical.
The solution, while hacky, is entirely workable. Esp. since the skull object will be small when rendered.
Create a small sphere INSIDE the skull and because its a different object, we can assign 0.0 darkening to it.
Whats this then?
For the last couple of days, I have been implementing a 3D web objects rendering project. Its eyecandy bling for my #grafana #prometheus stack.
Im nearing the end of the Minimum Viable Product (on the presentation layer).
I am 'coding' in #Javascript something I have familiarity of approximately 0.2.
So I thought it would be worthwhile to create some metrics from all the coding prompts.
The ratio of feature ready to buggy results is actually pretty decent.
It starts to get into the weeds towards the end, because I am developing 'featuritis' and pushing the initial spec.
Otherwise, the error to code ratio is comparable to the old school #programming but I suck as a #programmer regardless if its by hand or by #AI 🤡
N.B. The diagram is my invention, I might fiddle with it more to track #vibecoding efficacy and analytics
Do you crave cool?
Here it is.
Does not matter if you love #edm #techno or not...
@switchangel codes music line by line, her narration is part of this cool #trance track she magics line by line.
Whether you're a #codemonkey a #music #geek or just a #neurospicy unicorn
You will thank me for this gem.
Love it!
https://youtu.be/GWXCCBsOMSg?si=x2n3UonVu7qIhGw-
👨💻🚀 Oh look, they trapped an over-caffeinated code monkey in a while loop, and it birthed a half-dozen repos while the rest of us were busy dreaming about unicorns. 🦄 Because, who needs sleep when you have GitHub AI cranking out spaghetti code faster than you can say "debugging nightmare"? 🤖🍝
https://github.com/repomirrorhq/repomirror/blob/main/repomirror.md #codeMonkey #GitHubAI #spaghettiCode #debuggingNightmare #techHumor #HackerNews #ngated
Why Did a $10 Billion Startup Let Me Vibe-Code for Them—and Why Did I Love It?
I've gone back 40 years.
When I started out in computing in the 1980s it was common for programmers to carry around a utility library: a personal toolbox so you don't have to keep re-inventing your spanners.
That kind of stopped once programming languages' libraries got so big they practically encompassed everything.
/cont'd
#BostonWeekend 18/x
Fri #AimeeMann w/ #JonathanCoulton @ Wilbur, $80+
https://www.youtube.com/watch?v=F0V8CpT0Iow
https://www.youtube.com/watch?v=XgYT7zB3w5M
https://youtu.be/RSsstXfcRWw?t=65
https://www.ticketmaster.com/aimee-mann-22-lost-in-space-boston-massachusetts-06-06-2025/event/0100625AEA976BC0 #BostonMusic #BostonGeek #JoCo #CodeMonkey #Portal
What's an #MCP server?
(Model Context Protocol)
Today #Anthropic dropped V4 of their new model. It marks a pivot in their corp direction. They are yielding the ground for chatbots to #OpenAi and the rest, and instead focusing on infrastructure. Having used paid versions of #ChatGpt and #Claude for #vibecoding it's a smart move as Claude is significantly better than ChatGpt that's been lobotomised for minimum compute burn.
Back to MCP.
LLM --> MCP --> Any app
Eg. You can connect via the MCP to Gmail and issue a prompt "Send email to all the folks in my address list called Karen, Subject: Yo, Body: Your momma!"
Sure, you COULD do it with an API, but now peasants can use it, not just #codemonkey
#Zappier is the first MCP I know. There may be others.
Could AI program my programming? Maybe. But could it complain so bitterly about the solution that has to be implemented? I think not.
The bad part of troubleshooting computer programs is that sometimes part of it breaks, and then you have to suffer through people who want to be helpful asking if any other part of the computer program could have caused the problem, even stuff that doesn't actually touch it. To the point where it's like
Them: could it have been the microwave?
Me: be for real, James. How could it be the microwave?
And then sometimes it IS the fucking microwave. Like. Why.
It's wild to try to train someone to do your job and realize that, at the base, you are really trying to teach them how to think.
Also like don't tell anyone but I like helping with training the newbies because it makes it clear to me how far I've come. When I started this job I was barely past WRITE !, "HELLO CRUEL WORLD." Now I'm a pretty seasoned programmer.
I know 100% of things about the piece of the project I'm currently working on, and 0% of things about the piece I finished two weeks ago. Good thing I'm not in charge of—
Oh wait. 😐
Anyway, note to self to comment my code better.
Project architect: ...and the terminology project, which Em has been taking point on, has been very successful—
Me: (to myself) hang on, was I leading that? (Long pause) I guess I did all the analysis and development. Huh. Nice.
I gotta pay attention in meetings more.
Why is it that step 1 (put data into database) is always the most complicated?
Testing a thing. Update the API to return an error. Run the thing.
It returns an unexpected error.
Hmm. Back out changes. Try again. Unexpected error continues.
SO. Uh.
Boss: We need you to deploy this code to prod.
Me: This code? This code I wrote two months ago?
Boss: Yes.
Me: ...Remind me how it works again?
