Hacker News: steffs

New comment by steffs in "Giving AI Agents Database Access Is Way Harder Than It Looks"

steffs — Sun, 26 Apr 2026 00:11:19 +0000

The onion model is the right mental frame. The nastiest failures are often not obviously bad SQL, they are valid queries that become dangerous only after the planner sees real cardinalities. Row limits and statement timeouts help, but a query can still thrash caches or hold locks before timeout hits. Is your pre execution cost check based on an EXPLAIN style plan with relation level budgets, or is it mostly AST heuristics plus database backstops? That boundary usually decides whether something feels safe enough for production data.

New comment by steffs in "Headless Everything for Personal AI"

steffs — Mon, 20 Apr 2026 14:51:02 +0000

I think the deeper point is not CLI specifically, it is structured local tooling with predictable side effects. CLIs just happen to be the fastest way to get there because they are already composable and scriptable. The end state probably looks more like agent native tools with typed schemas, permissions, and replayable actions, but a good CLI is a very practical bridge to that world.

New comment by steffs in "Show HN: Clone, a small Rust VMM, forks VMs in under 20ms via CoW"

steffs — Mon, 20 Apr 2026 14:50:21 +0000

The part that stands out is that you are optimizing for warm state instead of cold boot. That feels right for dev shells. If the workload is repeated short lived environments, template fork time matters more than booting a minimal kernel fast. How do you handle template drift over time? Do you periodically rebuild and re-warm from scratch, or can you patch a warm template in place without losing the memory-sharing gains?

New comment by steffs in "Lemonade by AMD: a fast and open source local LLM server using GPU and NPU"

steffs — Thu, 02 Apr 2026 21:47:10 +0000

The multi-modal bundling is the part that stands out more than the raw inference speed. If you are building an app that needs text generation, image generation, and speech recognition, right now the local setup is three separate services with three different APIs and three different model management stories. Having one server handle all of that behind OpenAI-compatible endpoints is a real quality of life improvement for anyone prototyping locally. The NPU angle is interesting but probably overstated for most use cases. The discussion in the thread confirms what I would expect: NPUs shine for small always-on models and prefill offloading, not for the chatbot workloads most people care about. Where this gets genuinely compelling is if AMD can make the combined GPU plus NPU scheduling transparent enough that developers do not need to think about which hardware is running which part of the pipeline. That is not a solved problem on any platform yet, and if Lemonade gets it right for even a subset of workloads, it becomes the default choice on AMD hardware regardless of how it benchmarks against Ollama on pure text generation.

New comment by steffs in "Show HN: QuickBEAM – run JavaScript as supervised Erlang/OTP processes"

steffs — Mon, 30 Mar 2026 02:30:43 +0000

The no-JSON-boundary piece is the part that stands out to me. Most polyglot runtimes spend a lot of cycles serializing and deserializing at the language boundary, and that cost compounds fast when you are doing SSR or tight per-connection loops. Having Erlang read the native DOM directly without a string rendering step is a real architectural win, not just a convenience. Curious how you handle the supervision semantics when a JS runtime crashes.

New comment by steffs in "Wander – A tiny, decentralised tool to explore the small web"

steffs — Thu, 19 Mar 2026 14:19:35 +0000

The StumbleUpon comparison is apt but I think what made StumbleUpon work was the social layer: you could see what your friends upvoted, and that created an implicit filter against the pure randomness. Pure random discovery is fun for a session but gets old. Would love to see something like a lightweight trust graph here where a site vouching for other sites carries weight, similar to how Webring worked but with signal about quality rather than just affiliation.

Show HN: HiveSpace – GitLab Runner Health Dashboard

steffs — Wed, 21 Jan 2026 02:03:31 +0000

I spent hours last month debugging why GitLab CI jobs were stuck "pending" despite having active runners. The issue? Tag mismatches and an offline runner that wasn't obvious in GitLab's UI.

HiveSpace connects to your GitLab instance and shows:

- Why each stuck job isn't running (mismatched tags, offline runners, capacity) - Runner utilization and queue times at a glance - Trends to spot degrading performance

Built because GitLab's built-in runner metrics require Ultimate tier + ClickHouse setup, and many teams (including mine) don't have that. Also, even with those, the "why is this specific job stuck?" question isn't answered clearly.

Privacy note: Your GitLab API token is stored locally in your browser (cookie) and never sent to our servers. The dashboard makes API calls directly from your browser to your GitLab instance.

Early beta available. Book a demo at https://calendly.com/hello-hivespace/30min or just try it directly.

Stack: Next.js, GitLab API, TypeScript. Works with GitLab SaaS and self-hosted.

Comments URL: https://news.ycombinator.com/item?id=46700314

Points: 1

# Comments: 0

New comment by steffs in "Requiring 3D printers be equipped with certain blocking technologies"

steffs — Tue, 20 Jan 2026 03:17:48 +0000

Doesn't make sense... most of the software used on 3D printers is open source.

Show HN: Foom.ist: When silicon surpasses human brainpower

steffs — Tue, 20 Jan 2026 03:14:32 +0000

I built https://foom.ist, a real-time interactive visualization of when cumulative global chip FLOPS (since 1970) surpasses the cumulative "brain compute" of humans born since 1970.

You can tweak assumptions (brain FLOPS estimate, birth rate trend, Moore's law growth rate), zoom and pan the chart, and see the crossover date update live.

If you have better sources (birth rates, brain FLOPS estimates, chip production, compute), the site has a suggestion form that opens a GitHub issue. Code and issues: https://github.com/StefanSamne/foomist

Not selling anything. I mainly want feedback on the model and better data.

Comments URL: https://news.ycombinator.com/item?id=46687500

Points: 1

# Comments: 0

New comment by steffs in "What came first: the CNAME or the A record?"

steffs — Tue, 20 Jan 2026 02:29:23 +0000

I love this low-level stuff. Do you think this is the reason that in the past year or two DNS providers have stopped allowing CNAME records for the root domain? Meaning, it's to ensure backwards compatibility for certain DNS resolution tools that are order sensitive to not get confused?

New comment by steffs in "The saddest “just ship it” story (2020)"

steffs — Thu, 12 May 2022 14:55:07 +0000

The realness here is palpable

New comment by steffs in "Complexity Is the Mind Killer"

steffs — Tue, 10 May 2022 16:47:20 +0000

So guilty of this.... 1st thought "I should just learn this new skill so I can do it myself" 2nd thought "That will take days/weeks... learn enough to hire it out" 3rd thought "I'm glad I went with 2nd thought"

New comment by steffs in "Tell HN: Vitamin A Linked to Energy Levels"

steffs — Tue, 10 May 2022 16:44:19 +0000

100% been vegetarian for 10+ years and didn't realized I needed vit A suppl. till just a few years ago.

New comment by steffs in "Apple Discontinues iPod Touch"

steffs — Tue, 10 May 2022 16:42:46 +0000

The end of an era for sure.... hope the iOS keep coming out for as long as possible.

New comment by steffs in "Is launching on Product Hunt worth it?"

steffs — Sun, 08 Aug 2021 17:42:21 +0000

Great article. I'm curious, did you also post the startup on HN, and if you did, did you do it before or after PH?