Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
kstenerud
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
7 ms
·
61.
▲
by
kstenerud
2mo ago
The killer feature of golang for LLM dev is the tooling. forbidigo is what allows me to keep ambient config out of my app, and restrict file access to a small set of paths. The coverage tool has "nocover", so you can guarantee tha
62.
▲
by
kstenerud
2mo ago
Yup, and phoning a number wouldn't incur charges unless the other side picked up. ringconnectd never answered a call, but rather listened for a ring pattern that you could specify. So what I'd typically do is go to a payphone, dia
63.
▲
by
kstenerud
2mo ago
I was pissed when I discovered that the cassette tape containing all of my c64 programs was gone. Most of them were crappy, but there was a D&D character generator I wrote when I was 8 that I'd made extensive use of for quickly gen
64.
▲
by
kstenerud
2mo ago
I wrote one that has both of those: yoloAI (MIT, Go, single binary, no login). https://github.com/kstenerud/yoloai Outbound firewall is `--network-isolated`: egress is denied except the agent's own API endpoints p
65.
▲
by
kstenerud
2mo ago
> We can all agree: the incentives are bad. No, we don't. > Everywhere you look, people are being rewarded for doing the wrong thing. And, people are also being rewarded for doing the right thing. Don't conflate the two. Jus
66.
▲
by
kstenerud
2mo ago
I've always taken it to mean "coding is the easiest part of a developer's job" That doesn't make it inherently easy, but it is the easiest part to automate. A modern LLM is perfectly capable of maintaining decent co
67.
▲
by
kstenerud
2mo ago
Nope. For me the constraint is the same as before: time. A 2-day vibe coded project is a prototype, not a product. An actual product takes months. AI merely allows you to build more ambitious products.
68.
▲
by
kstenerud
2mo ago
Yes, I was up at 4 am every day. It's hard work.
69.
▲
by
kstenerud
2mo ago
The author talks as if this were a new phenomenon; it is not. I actually did go buy a farm in the 2010s, grew alfalfa, managed cattle, cared for horses, etc. Except I did it while working remote as a software engineer. Very fun life, and I
70.
▲
by
kstenerud
2mo ago
Permission prompts is a TERRIBLE model, and never should have existed. This is one of the reasons that led to the development of yoloAI: - No permission prompts. The agent has free reign and never has to ask permission, but is in a sandbox.
71.
▲
by
kstenerud
2mo ago
You could actually use SIMD instructions to detect sequences of bytes with the top bit cleared, thus allowing bulk copies of ASCII text without a per-byte loop. Another optimization would be to take advantage of the fact that codepoint usag
72.
▲
by
kstenerud
2mo ago
Just like published books were considered a symbol of effort that no one would have considered a blog as being anywhere near. When was the last time you looked up a book to find an answer to your question? That USED to be the normal. Then w
73.
▲
by
kstenerud
2mo ago
These are all symptoms of the adjustment period following a disruptive technology. Eventually it'll be perfectly normal to use AI-generated content (and use AI to summarize it), and nobody will bat an eye.
74.
▲
by
kstenerud
2mo ago
Ultimately, the problem is that every person has their own anecdata-driven or follow-the-leader-inspired ideas about how to interview effectively, and large swaths of them are in stark opposition to each other. And now after decades of rese
75.
▲
by
kstenerud
3mo ago
I think that the author is focusing on the symptom rather than the cause. The current nerd bro culture isn't much different from the jock bro or wall street bro or adventurer bro cultures that preceded it. That it wears clothing famili
76.
▲
by
kstenerud
3mo ago
... And then drive under the threat that their own tires will get punctured as well.
77.
▲
by
kstenerud
3mo ago
That would be an expensive way to drum up business.
78.
▲
by
kstenerud
3mo ago
This sounds a lot like: "Try doing this thing." "No." "You can't do it because of your deficiencies." "I won't do it because all you've done is spout a bunch of pseudoscience."
79.
▲
by
kstenerud
3mo ago
It's why I built the yoloAI tool, in fact!
80.
▲
by
kstenerud
3mo ago
Also known as the Buttle-Tuttle mixup.
81.
▲
by
kstenerud
3mo ago
> That same time-traveling video renter might find herself startled by the employee offering to recommend a movie, or by the knot of customers hanging out by the counter, shooting the breeze and discussing film. What universe is this per
82.
▲
by
kstenerud
3mo ago
I built yoloAI, which is a sandboxer so it uses tmux to keep the agent alive inside. Since I can just reattach to them or use VS code tunnels, I don't really have a need for a multiplexer, though.
83.
▲
by
kstenerud
3mo ago
I've been dealing with it on the post-training (runtime) side with a large codebase that contains many moving parts, many rules and requirements. Putting all of that into the context has already gone far past 1MB, so it's untenabl
84.
▲
by
kstenerud
3mo ago
Did you look at the starling code?
85.
▲
by
kstenerud
3mo ago
You won't, because your prejudice will prevent you from looking.
86.
▲
by
kstenerud
3mo ago
Nicely done! This is an example of how engineering discipline mixed with good agent guidance can avoid the slop. A couple of things: - The test pyramid (static -> unit -> functional -> VM release gate) works but is dependent on som
87.
▲
by
kstenerud
3mo ago
> It might seem absurd to you to even suggest superhuman AIs could function as a truth-oracle (it certainly does to me), but there are two reasons to take it seriously. First, it is how these things will be used practically by the vast m
88.
▲
by
kstenerud
3mo ago
Go and rust have better guardrails that help agents write better code. Python and JS aren't opinionated enough.
89.
▲
by
kstenerud
3mo ago
A toilet's operation wouldn't be too hard to explain to an alien who has enough science for interstellar travel. A siphon mechanism works the same way for them as for us (although they'd likely wonder at how primitive we are
90.
▲
by
kstenerud
3mo ago
I'm not doing adversarial critiques of the code; I'm doing adversarial critiques of the designs and plans. If it's still outputting bad code when it comes to the implementation, I do an interrogation round to find out why it&
More ›