HNHacker News
TopNewBestAskShowJobs

tuhgdetzhh

239 karma · joined October 18, 2025

submissionscomments
tuhgdetzhh··on Iran Goes Into IPv6 Blackout
Isn't it possible to jam the starlink receiver?
tuhgdetzhh··on LaTeX Coffee Stains (2021) [pdf]
"This page was intentionally left blank" is also an all time favorite of mine.
tuhgdetzhh··on Approaching 50 Years of String Theory
An in half a century we still haven't found a single actually testable prediction.
tuhgdetzhh··on From Zero to QED: An informal introduction to formality with Lean 4
Lets assume the kenerl is correct and a mathematian trusts it, now what happens if it is the proof itself that he cannot understand anymore because it has gotten too sophisticated? I think that this would indeed be a difference, and a crossing point in history of mathematical proofs.
tuhgdetzhh··on We pwned X, Vercel, Cursor, and Discord through a supply-chain attack
Could you elaborate on why not?
tuhgdetzhh··on Carrier Landing in Top Gun for the NES
Just for comparison, this is how the code could look like in Python:

  SUCCESS = 0
  TOO_FAR_LEFT = 2
  TOO_SLOW_OR_TOO_LOW = 4
  TOO_FAST_OR_TOO_HIGH = 8
  MIN_ALTITUDE = 100
  MAX_ALTITUDE = 300
  MIN_SPEED = 200
  MAX_SPEED = 400
  MIN_SPEED_200_RANGE = 238
  MAX_SPEED_300_RANGE = 338
  MAX_HEADING_RIGHT = 8

  def landing_skill_check(
    altitude: int,
    speed: int,
    heading: int) -> int:

    if altitude < MIN_ALTITUDE:
        return TOO_SLOW_OR_TOO_LOW
    if altitude >= MAX_ALTITUDE:
        return TOO_FAST_OR_TOO_HIGH
    if speed < MIN_SPEED:
        return TOO_SLOW_OR_TOO_LOW
    if speed >= MAX_SPEED:
        return TOO_FAST_OR_TOO_HIGH
    if speed < 300:
        if speed < MIN_SPEED_200_RANGE:
            return TOO_SLOW_OR_TOO_LOW
    else:
        if speed >= MAX_SPEED_300_RANGE:
            return TOO_FAST_OR_TOO_HIGH
    if heading < 0:
        return TOO_FAR_LEFT
    if heading >= MAX_HEADING_RIGHT:
        return TOO_SLOW_OR_TOO_LOW

    return SUCCESS
tuhgdetzhh··on Python Workers redux: fast cold starts, packages, and a uv-first workflow
Even if you actually use the network module in Go, just so that the compiler wouldn't strip it away, you would still have a startup latency in Go way below 25 ms from my experience with writing CLI tools.

Whereas with Python, even in the latest version, you're already looking at atleast 10x the amount of startup latency in practice.

Note: This is excluding the actual time that is made for the network call, which can of course also add quiete some milliseconds, depending on how far on planet earth your destination is.

tuhgdetzhh··on Is your AI system illegal in the EU?
The answers is: likely yes, but the solution is to become the market leader now and pay the fines later. This buisness strategy has worked out very well for the magnificent 7.
tuhgdetzhh··on AI should only run as fast as we can catch up
> One problem is that the idea of being "well-structured" has gone overboard at some point over the past 20 years

That's exactly my experience. While a well-structured monolith is a good idea in theory, and I'm sure such examples exist in practice, that has never been the case in any of my jobs. Friends working at other companies report similar experiences.

tuhgdetzhh··on AI should only run as fast as we can catch up
Yes, unfortunately those who jumped on the microservices hype train over the past 15 years or so are now getting the benefits of Claude Code, since their entire codebases fits into the context window of Sonnet/Opus and can be "understood" by the LLM to generate useful code.

This is not the case for most monoliths, unless they are structured into LLM-friendly components that resemble patterns the models have seen millions of times in their training data, such as React components.

tuhgdetzhh··on Java Hello World, LLVM Edition
It would raise the same kind of alert for me if someone used wget to download a binary executable instead of a shell script.

The issue is not the specific form in which code is executed on your machine, but rather who is allowed by you to run code on your computer.

I don't trust arbitrary websites from the Internet, especially when they are not cryptographically protected against malicious tampering.

However, I do trust, for instance, the Debian maintainers, as I believe they have thoroughly vetted and tested the executables they distribute, with a cryptographic signature, to millions of users worldwide.

tuhgdetzhh··on Java Hello World, LLVM Edition
I'm always a bit shocked how casual people people wget and execute shell scripts as part of their install process.

This is the equivalent of giving an author of a website remote code execution (RCE) on your computer.

I get the idea that you can download the script first and carefully read it, but I think that 99% of people won't.

tuhgdetzhh··on The unexpected effectiveness of one-shot decompilation with Claude
But on the other hand, at the current speed of LLM progression, a game that might have been obfuscated with the help of Opus 4.5 might in two years be decompiled within hours by Opus 6.5.
tuhgdetzhh··on Free Nano Banana Pro
But why posting a "free" product that becomes unfree after a hug from HN, so that the vast majority will see an unfree product and thinking you were baiting them with false claims?

EDIT:

> Due to this incredible demand, we've hit our current budget limit and need to temporarily pause the service.

There you go.

tuhgdetzhh··on 100k TPS over a billion rows: the unreasonable effectiveness of SQLite
Or rent a bare-metal machine from hetzner with 2-3x performance per core and 90% less costs[1].

[1] Various HN posts regarding Hetzner vs AWS in terms of costs and perf.

tuhgdetzhh··on A Love Letter to FreeBSD
Yeah, sorry for removing that part. Changed my mind just minutes after posting, because I really like FreeBSD any my critique sounded a bit too harsh.
tuhgdetzhh··on A Love Letter to FreeBSD
A love letter to the last operating system that isn’t trying to gaslight you. FreeBSD really is the anti-hype choice: no mascot-as-a-service, no quarterly identity crisis, just a system that quietly works until the heat death of the universe.
tuhgdetzhh··on Steam censors LGBTQ+ content on behalf of the Russian Government
For all of you who played Wolfenstein 3D in your youth in Germany, here is an update: It is now completely legal since 2019.
tuhgdetzhh··on TPUs vs. GPUs and why Google is positioned to win AI race in the long term
What is the difference between a DSP and Asic? Is a GPU a DSP?
tuhgdetzhh··on What OpenAI did when ChatGPT users lost touch with reality
Your comment has too many em-dashes for my taste.
tuhgdetzhh··on Cloudflare Dashboard and Cloudflare API service issues
Hopefully not again.
tuhgdetzhh··on Do not put your site behind Cloudflare if you don't need to
Yeah, but I need it.
tuhgdetzhh··on Cursor: Past, Present, and Future
They propably burn something in the order of 50M-100M per month in LLM API costs for models like Sonnet 4.5. So the answer would be: Yes.
tuhgdetzhh··on We cut our Mongo DB costs by 90% by moving to Hetzner
I recently did a total cost of ownership analysis for moving off AWS to Hetzner: https://beuke.org/hetzner-aws/
tuhgdetzhh··on Yt-dlp: External JavaScript runtime now required for full YouTube support
Only if you use Googles Chrome Browser. Other browsers have issues like Firefox[1].

[1]: https://news.ycombinator.com/item?id=41379517

tuhgdetzhh··on Waymo robotaxis are now giving rides on freeways in LA, SF and Phoenix
If he can create a buisness of operating a fleet of self driving cars fine, but 99% of regular taxi/uber drivers will loose their job.
tuhgdetzhh··on Modular monolith and microservices: Modularity is what matters
Modular monolith is a pipe dream. Great idea, but it gets messed up in reality (similar to communism). You just need a bit of time pressure and project deadlines and the devs will crack and throw the modularity over board with just a few merges. Multiply that over a decade, then you end up with a monster and fight dragons.
tuhgdetzhh··on ClickHouse acquires LibreChat, open-source AI chat platform
Yup, they will of course try to profit long-term, ClickHouse is not a charity project.

Within the next few years, it will introduce an additional enterprise edition, a SaaS offering, or a change in licensing terms.

tuhgdetzhh··on ClickHouse acquires LibreChat, open-source AI chat platform
History repeats itself. I wouldn't be suprised.
tuhgdetzhh··on Asus Ascent GX10
For comparison, the RTX 5090 has a memory bandwidth of 1,792 GB/s. The GX10 will likely be quite disappointing in terms of tokens per second and therefore not well suited for real-time interaction with a state-of-the-art large language model or as a coding assistant.
← PreviousPage 2 of 3Next →