@kerckhove_ts Will give it a look.
Gitlab with to-be-continuous and a common pipeline has been a solid pattern for me. Argo and zarf for CD.
English
Systems Pimp ๐ฉ๐ฆฏ๐ถ๏ธ
2.7K posts
Systems Pimp ๐ฉ๐ฆฏ๐ถ๏ธ
@Hunters_laptp
Exploiting bad design with its own rules. Turning broken systems against themselves.
๊ฐ์
์ผ Mayฤฑs 2023
83 ํ๋ก์98 ํ๋ก์
@Hunters_laptp Config-free CI :)
And Nix for CD
See nix-ci.com
English
This is how I feel about YAML configuration for CI jobs .
taoki@justalexoki
this is probably my worst tweet but i honestly can't wait for our cats to die. ever since we got our kid they've been nothing but a nuisance. all they do is make a mess and cost money. i just don't care for them at all anymore. it's weird
English
@slimjimmy I out grew my mental mutilation by high school when I found Turbo PASCAL.
English
"It is practically impossible to teach good programming to students that have had a prior exposure to BASIC: as potential programmers they are mentally mutilated beyond hope of regeneration"
- Dijkstra
alcuin โ๏ธ@scheminglunatic
i dont remember who said that thing recently that computer programming makes u think more logically but i think its true but i think like if u use a bad langauge it permanently deforms ur brain and thats why software people are so fucked up and wrong about everything
English
@techspence Chinese models are 100% a no go in defense for sure.
English
@systemdesignone But it worked on my machine.
Rootless Podman is the way.
English
Facts about Docker
>Adds unnecessary overhead
>Makes networking harder
>Introduces security risks
>Turns simple apps into complex ones
>Increases operational complexity at scale
>Makes debugging difficult
>Bloats with inefficient images
Why do people like this thing?
English
@JeffBohren Most of these points are mitigated by mature DevSecOps practices.
English
I fear giving AI agents access to sensitive information and systems is going to be one of the worst security mistakes IT has made recently.
This for at least three reasons:
1) LLMs are vulnerable to prompt attacks. Further, no one seems to know how to prevent it.
2) MCPs will continue to be a source of RCE vulnerabilities. It reminds me of the CGI-Bin days of the internet.
3) Skills will continue to be vulnerable to supply chain attacks.
Did I leave anything out?
English
@heygurisingh Been treating it like a jr the whole time.
English
Vibe coders are not going to like this.
UC San Diego just published the first real field study of experienced developers using AI agents. They watched 13 of them code in the wild and surveyed 99 more.
Zero of them vibe coded.
Not one developer "fully gave in to the vibes." Not one trusted the agent to ship. The researchers found the opposite of what every Cursor demo on your timeline implies.
Experienced devs plan before they prompt. They load the agent with heavy context. They verify every diff and refuse to merge code they haven't actually read.
"Flow and joy" coding, the whole Karpathy vibe coding pitch, got quietly rejected by every professional in the study. They said it's fine for throwaway prototypes. Not for anything that ships.
The devs still liked using agents. They just don't let the agent drive.
Turns out the people who've shipped software for a decade know something the vibe coding influencers don't.
Huang et al., UC San Diego. December 2025.
Paper in comments.
English
@crochet_mom314 We had several on the educators track accidentally sign up for CompOrg thinking it was an entry level productivity class. It was comical and lead us to discover all the dumb dumb classes for them.
English
As someone who got a teacher's certification alongside my degree in chemistry, the majority of you are ABSOLUTELY smart enough to homeschool your kids. The education classes I took were by far my easiest courses in college.
English
@sudoingX Coder and gitlab with to-be-contenuous for the win.
English
watching the replies on this post is more useful than the post itself. nobody disagreed. every reply was some flavor of "i forgot," "i put it aside," "not excited to set it up."
this wasn't an unpopular opinion. it's the universal thing builders know but defer because setup is boring. the defer pile is where the moats hide, every finished one becomes a leverage point. builders already know. doing is the hard part.
Sudo su@sudoingX
unpopular opinion anon: if you don't have your own private git server you probably won't make it.
English
@UKBelle99 @DarrigoMelanie Proving citizenship vs age are wildly different. You don't need to be a citizen to drink at a bar. You do need to be a citizen to vote.
English
@Hunters_laptp @DarrigoMelanie What did she say that was wrong? Unless you're in a handful of states that identifies citizenship on the Real ID, your Real ID won't work at the voting booth. But it will work at a bar.
English
The ID you use to enter your bar is not an acceptable ID to vote under the SAVE Act.
Try only accepting SAVE Act-approved ID and see how fast you go out of business.
Jimmyโs Famous Seafood@JimmysSeafood
You need valid ID to enter our bar. This isnโt a voting booth.
English
@AndrewYang Wildly unrealistic. You probably mean to tax inference beyond the sales tax that already occurs? What of local inference? Additionally why strangle one of the few economic drivers atm?
English
We should tax the bots. blog.andrewyang.com/p/tax-the-bots
English
@TechOperator FWIW FSD makes it easier to chamber a round when needed.
English
Six million people will crash while driving this year. But sure, you are a good driver, and no, you donโt need a self-driving car.
TechOperator@TechOperator
โI just took my eyes off the road for a second.โ Thatโs the classic line people say right after a distracted driving moment causes a crash, and itโs one of the most common excuses. Most people are good, safe drivers when paying attention. But letโs be honestโฆ - Have you ever taken both hands off the wheel to open a drink? - Have you taken your eyes off the road because of a spill? - Have you had to quickly avert your eyes to your children to see what the ruckus is about? - What about those glances at your phone notifications? - Or that quick text to let someone know youโll be late? Letโs be honestโฆ You have done things like this while driving, or you are lying. Over the past 5 years (2020โ2024), the average is about 5.9-6.1 million police-reported crashes per year. But sure, you are a good driver. No, you donโt need a self-driving car. Still, 6 million of you will crash this year. And thatโs the power of Tesla Self-Driving: momentarily engage to deal with whatever and stay safe. Or better yet, just let it handle the driving for you.
English
@Hamdogg @DarrigoMelanie You know what they say about assumptions.
English
@Hunters_laptp @DarrigoMelanie I assume you are talking to the restaurant
English
@BasedMikeLee Or here me out. The House and Senate could get back to that pesky line item balanced budget process we used to follow. DOGE is only needed because we haven't had a real budget since 96.
English
The national debt has grown faster than the U.S. economy for decades
We need DOGE 2.0
And we need Congress to stop spending nearly $2 trillion a year more than it brings in
Wall Street Mav@WallStreetMav
Our govt national debt, compared to the size of the US economy, is racing into insane territory. DOGE was the last chance to change the trajectory. We will see this system collapse within our lifetime. The interest payments on the debt are swallowing the bulk of the budget.
English
@Logically_JC No easy botton exists and pretending otherwise is intellectual masterbation.
English
So many of our problems could be fixed by properly funding public education.
English
@spinelessaisha Check out Podman or if not on Windows rootless Podman.
English
@mattpocockuk I built a harness where the model manages it's context on it's own as well as reworking how tools work. It lets your find the ceiling of a model.
English
What 'advanced' AI coding techniques are you using?
I.e. what do you feel like you've discovered that no-one else knows about yet?
English
@iamtrask Air gapped deployments and edge compute. On-boarding, documentation, testing, pr review, insights into DSU and sprint goal alignment, just to name a few.
English
Does anyone have a Local AI setup which they are *convinced* is... like... actually better than Cloud AI for some usecase?... and why? (I assume... data too private to share or latency or something) I'm trying to get a better sense of where the perceived value is these days.
โฟป Andrew Trask@iamtrask
Hey - if you donโt use a local LLM as your daily driver, why? What are the remaining problems?
English
It's all about the harness!
My harness is tiny ~500 token system prompt including tool defs.
Making the model aware of it's state changes everything.
How it started: x.com/Hunters_laptp/โฆ
Systems Pimp ๐ฉ๐ฆฏ๐ถ๏ธ@Hunters_laptp
English