Marek Suscak

Choosing the right data structure can make a big difference. Bloom Filters, Hashtables and Tries (Prefix trees); I've created and compared 3 different implementations of the spell checker. @cs50 @Harvard @davidjmalan

@mariorod1 And then there's a few-second lag after a PR gets merged and before it's actually removed from a filtered list. It's probably an acceptable tradeoff, but it can be misleading.

@mariorod1 I don't mean to rub salt in the wound, but this Elasticsearch implementation has not been great since the get-go. I had to reach out to support a while back to have them rebuild an index in some of our repositories because they had somehow gotten out of sync with your DB.

After yesterday’s incident, we are investigating cases where /pulls and /repo/pulls pages are not showing all indexed pull requests. This is because our Elasticsearch cluster does not currently contain all indexed documents. No pull request data has been lost. As pull requests are updated, they will be reindexed. We are also working on accelerating a full reindex so these pages return complete results again. We are statusing PRs degraded to reflect the remediation: githubstatus.com/incidents/x69z…

Its helpful to remember that Dario saying this is approximately no different than any random person off the street. He does not have the expertise for many of the claims he makes and frankly should stop sharing his irrational opinions publicly. Its classic FUD (Fear, Uncertainty, and Doubt). Its purely damaging for Anthropic at this point.

miseの作者が作った新しいNode.jsパッケージマネージャー。package-lockの以降が不要でかつ他より高速。デフォルトでminimumReleaseAge=1440が設定されておりセキュリティにも配慮されている。 / “aube” htn.to/4BVK18cQzr

Mozilla says Mythos helped identify 271 vulnerabilities in Firefox 150. I went through the commits, CVEs, and bug links to see what that number really means. My takeaway: relax folks. xark.es/b/mythos-firef…

There's a really high cost of duplicating code and we're all relearning that everyday with these coding agents. Until they get better at generating reusable code, bugs will run rampant. You can't "abstract it away" with more agents. You just get crappy software....

“Don’t buy a car, it’s a depreciating asset.” Brother, I myself am a depreciating asset. I won’t be here forever. Not everything has to be optimized for returns, some things are just meant to be enjoyed while we’re here.

we knew how to make bad code, cheap, and fast, before agentic coding emerged. why didn't we follow that path earlier? it bugs me. why now? we had the knowledge how to build bad code before.

You have to understand. Spending 1 hour per day on brainrot is insane. That's about 6% of your waking day. About 5 years of your waking life. Half a decade. On brainrot. Just gone. Zero return. Zero fulfillment. Zero meaning. Zero contribution to the other parts of your life.

I’m starting to think software engineering won’t die, it will just expand in role responsibilities as coding shrinks. You will be a software engineer + security engineer + data engineer + AI/BI engineer all rolled into a single role. That is a minimal expectation.

Never agree to carry pressure in a situation where you have no right to shape the outcome, because responsibility without authority does not make you important for long. It makes you absorb blame cheaply while other people enjoy the comfort of your obedience.

In an AI-driven world, an engineer who can write 10x more code is less valuable than a leader who can decide which 90% of that code shouldn’t be written.

I'm also skeptical about AI reducing demand for skilled engineers so far. For most companies without an overhiring problem AI codegen has led to an *increase* in engineering demand to keep up with the tech boom and all the new competitors on the market. If you're cutting staff instead of doing more things that's not a great sign.

in times when most people seem to only care about shortcuts, this is a nice thing to read

@eatonphil Did you read the 1st edition by any chance? Is it worth getting the 2nd one if I read and still own the 1st edition?

I first tried to read this book in 2018 and couldn't make it through because I thought it was too hard. 8 years later it's the only book I recommend every developer reads, and I had the chance to review the 2nd edition. Join a study group, give it a read.

I have 1000 ideas I've never built because I always get stuck thinking about how to build them right. Meanwhile, every week I watch people who don't know what a database is ship working products. This is eye-opening. Sometimes, code quality, principles, theoretical concepts, and all of that matter. Sometimes, it just gets in the way.

@tpschmidt_ What are the DOs and DON’Ts when using self-managed Karpenter (i.e. not EKS Autopilot)? Any common failures to watch for?

It makes the infrastructure feel (almost) invisible and invincible! Devs just request resources in their manifests and the compute appears 🥰

Karpenter is great because it ignores the idea of groups entirely. It looks at exactly what your pods need and talks to the AWS API to spin up the right instance at that moment 🏗️ The process is straightforward: 1. A pod can't find a home

generalists are about to win big If you understand a little of tech, business, and people, and can connect everything fast. you're sitting on a goldmine right now.

You still have to refactor, even when you use AI Agents. What's different is the way you discover that need, and the speed with which it occurs: adamhjk.com/blog/you-still…

Many don't get why this is cool. I spent a lot of time optimizing web performance. It was one of the main things I did at both lovable and spawn. Here's somewhat simple and thorough explanation: The browser rendering pipeline goes: Style → Layout → Paint → Composite. Layout is the expensive step. It calculates where everything goes on the page. Positions. Sizes. Line breaks. When you read offsetHeight or getBoundingClientRect, you force the browser to run layout. If you wrote to the DOM before that read, it recalculates everything. That's reflow. A lot of layout work is text measurement. Where does this line wrap. How tall is this paragraph. How wide is this word. The browser is the only thing that can answer these questions. There's no alternative. You have to go through the DOM. This is the root cause behind a lot of performance problems people deal with daily. Virtual scrolling exists because you can't know element heights without rendering them first. CSS contain exists to tell the browser "don't recalculate layout outside this box." Batching DOM reads and writes exists because mixing them causes layout thrashing. Text measurement is a big part of all of these. It's slow and locked inside the browser. You see it in real apps too. Slack estimates message heights for virtual scrolling. When the estimate is wrong the scroll jumps. You've seen this. Google Docs recalculates every paragraph below your cursor when a line wraps. Every keystroke. This is why it gets slow on long documents. AI chat apps get janky when streaming because each new token can cause a line wrap which changes the height which causes the page to jump. Same problem every time. Need text measurements. Only way to get them is the DOM. DOM is slow. --- Over the years people have moved pieces of UI work out of the browser's layout engine. Rendering moved to Canvas and WebGL. Scrolling moved to custom implementations with transforms. Positioning moved to JavaScript-calculated coordinates. Interaction handling was always just JS. But text was the one thing you couldn't move out. You always had to go back to the DOM to ask "how does this text wrap." Every other piece had an alternative. Text didn't. --- Enter Pretext. Pure TypeScript text measurement and line wrapping. No DOM. No reflow. You give it text, a font, and a width. It returns exact line breaks, widths, heights. Just a function call. Pure math. Text measurement was the last piece of layout with no alternative outside the DOM. Now it does.