you know that saying about how a compiler made by N teams will contain N+1 passes?
like every big program contains a replica of the organizational structure that created it?

It's fun sometimes to see big programs made by one person, because they're a sort of inverse mold of their brain structure.

linus torvalds said something like "every project has to fit within one person's brain" but sometimes the opposite is true
anyway the minor problem I'm having (this tool doesn't properly handle unicode filenames despite having unicode support) is looking to be so intractable that my options are either:
1. find and switch to a completely different tool that doesn't have this problem
2. find the programmer and eat their brain, gaining their knowledge of how their code works. then implement a fix

arg. it seems this tool is so dominant in this space (because it does Basically Everything) that there is no real alternative.

so it's back to the cannibalism, I guess

@foone Which tool?
@foone That was not intentionally with two interpretations, but if you can tell us the tool that wrote that tool that'd be good too 😄
@dascandy42 Wiimms ISO Tools.
@foone Looks to be completely on Github. Can't find any obviously wrong utf8 code. Perhaps you can find the problem in the code? https://github.com/Wiimm/wiimms-iso-tools
GitHub - Wiimm/wiimms-iso-tools: »Wiimms ISO Tools« is a set of command line tools to extract, modify and create Wii and GameCube ISO images and WBFS containers.

»Wiimms ISO Tools« is a set of command line tools to extract, modify and create Wii and GameCube ISO images and WBFS containers. - GitHub - Wiimm/wiimms-iso-tools: »Wiimms ISO Tools« is a set of co...

GitHub
@foone you could also bodge it to work well enough for your usecase (I think it was SHIFT-JIF not working right?). Bit of a pain but. Less to understand.

Edit: Welp, from the other thread that seems to be what you're gonna do so...

@foone

Definitely #2.

I mean, that's just good science.

@foone have you tried any LLM tools to analyze the code base? I haven’t seen one that performs better than grep, but an “inverse mold of someone’s brain” might be an interesting test case.

@foone Regarding option 2: MIND FLAYER! YAY!!!

(Just being silly...)

@foone 🚀 rewrite 🚀it🚀in🚀Rust🚀
@foone every… brain has to fit in one person’s project?
@demize I meant more "every project has one person's brain in it" but sure. that too
@foone
"Every person's brain has to fit within one project" sounds too matrix-y for my taste
@RedstoneLP2 get a load of this loser who still wants to be tethered to The Flesh
@foone who said anything about being tethered to The Flesh? i just don't want to share a repo with some people...
@foone @RedstoneLP2 "the project hit its MVP milestone and is shipping so we need to simplify your brain to match, we'll circle back and re-evaluate stretch goals" is the horror case there

@foone

I have been told that my code is “unusually focused on static lookup tables” which I guess says something about my brain. I do love big tables and small functions.

@lorddimwit @foone I always prefer table driven code over N slightly modified repeated statements.

@lorddimwit I love that Cake song!

I want a guy with a short function and a looooong (lookup) table!

@lorddimwit also, same. Although I tend to go with lookup dictionaries when I'm programming in Python, which is most of the time. But when I can't, it's lookup tables.

You can really see how I tried very hard to build a data-driven vector drawing language in a limited form of C++ here, and how much of a pain it ended up being:
https://github.com/foone/VGAPride/blob/main/flagdefs.cpp#L476

GitHub - foone/VGAPride: An MS-DOS application for showing Pride Flags

An MS-DOS application for showing Pride Flags. Contribute to foone/VGAPride development by creating an account on GitHub.

GitHub

@lorddimwit @foone oh, damn, now that I think of it... same!

My most recent project even has a whole file that's just a bunch of lookup tables.

It just seems much nicer than endless switches, especially when the same variants come up in several places. Keeps the actual code short and neat.

[mentionuserid="6c33d3d7-98d8-41d8-bad5-d2b2896a4052"] please take a look into my brain

It started as a tumblr clone and... well... here we are.

I think I might had spamed it before it had fediverse integration


#ap-was-hell#ap-is-still-hell
GitHub - gabboman/wafrn-backend

Contribute to gabboman/wafrn-backend development by creating an account on GitHub.

GitHub
@foone npm 1 was a case study in this. So was early nodejs
@foone I think large part of this is just that understanding code is often really hard, even and often especially, if it's well optimised and efficient. Good documentation can help greatly here, but it's often missing and few projects really maintain a detailed one.