i’m trying my best to summarize my dislike of 3d printing because it comes in waves. it boils down to: it’s not a serious process because of simple physics on one hand, it is a deeply useful technology for non-structural prototyping. but so are hammers and neither are revolutionary enough to “bring manufacturing back to the united states” it’s anisotropic. e.g. you can’t analyze it to determine requisite part strength against things like yield or ultimate stress it’s slow - andrew mccallip showed you can CNC parts faster than 3d printing them recently (link below) it’s susceptible to cold flow (creep) in a way that makes injection molding look like 4340 the unit economics never make sense past onesie-twosies, even if the physical properties weren’t a blocker it isn’t water tight (i mean, come on) but my biggest complaint is that it inspires lazy thinking. we live in the most exciting time ever for manufacturing. i can click a buy button on a site like sendcutsend and get production grade aluminum or steel parts in days note that i didn’t complain about strength - that’s the least of its problems anyways, to summarize, get better at mechanical design and design with real materials so we can truly reindustrialize. thanks for coming to my ted talk

Sad to see a generation of Linux users not understanding a pre-systemd world. Shell scripts are what made UNIX. Please, return to them. It was a more user friendly world.

i would simply declare that my computer has firefox on it and then re-create the entire universe

Imagine IDs are like toy boxes labeled with the time you got them. Some say this clumps new toys together, making a "hot" mess in one spot. But that's good! Toys from the same playtime are nearby, so you grab them fast without searching everywhere. That's "locality of reference" – like keeping your favorite toys close for quick fun.

you’re gonna love this openjdk.org/jeps/526

I didn't have the guts to do this so I invented caddy's plugin system

I can respect the hustle of being so averse to functional programming that you write 40 long ass bash scripts just to avoid doing a single file of it

They're reinventing NixOS from first principles

Much of the the internet runs on Elastic Block Storage. It's the default (and in most cases, required) storage layer for every EC2 instance running in AWS. This article by @molson was a delightful read on its history and engineering challenges.

i am once again reminding you to think about your data and access patterns

Virginia voters implement a PID control loop but the integral gain is too high

man how can you let sponsored results take up the whole screen

living the dream

The magical thing about D&D is that 1/20 is just barely under the threshold at which we can reason intuitively about probabilities. 1/4? I know what this is, intuitively. 1/10? Yeah, I see that all the time. 1/20? This is now a Small Probability according to the brain. Two nat 20s at the same time? That's two Small Probabilities at once! Nearly impossible. But 1/400 is actually quite likely if you consider the number of times you roll a D20 over the course of a D&D campaign.

Every Microsoft Entra tenant is weird in some way, but every IT admin is convinced they’re totally normal. We have thousands of customers at Pistachio, and we still meet admins who surprise us. “Of course we put company name in the email field and email in department”

Fun fact: at one point Google had an entire team building a new sandboxing technology just so they could run ffmpeg safely. Later it ended up being used in App Engine and other environments.

RDS: “You’ll put up with it because databases are scary.”

Counter-intuitive advice for designing very large scale live-site services: 1. Don't have retries inside your system, only at the edges. 2. Don't have queues inside your system, only at the edges. 3. When shit really goes down process requests in LIFO order.

Once a queue in a distributed system or a buffer in a network gets large enough it takes so long to reach the front that the client has already timed out and retried. You'll spend all your time processing useless work. Good systems have small queues. If your queue has to get long then switching to LIFO under load at least ensures you're making progress on requests users still care about.