Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
esafak
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
26 ms
·
751.
▲
by
esafak
6mo ago
What did you love about this show?
752.
▲
by
esafak
6mo ago
It should be understood that he canceled before Google started charging him.
753.
▲
by
esafak
6mo ago
It's been said before, but it is important to prospective users, so it bears repeating: screenshots and benchmarks, please; it helps users decide whether to invest time in it. The ability to transfer settings from other agents would be
754.
▲
by
esafak
6mo ago
Why did Svelte affect your stance towards Tailwind?
755.
▲
by
esafak
6mo ago
Put them in a robot so that it can navigate the physical world like humans. Self-driving cars.
756.
▲
by
esafak
6mo ago
In http://lobste.rs/ the lobsters are us :)
757.
▲
by
esafak
6mo ago
Apparently it does now: https://packaging.python.org/en/latest/specifications/pylock... https://pip.pypa.io/en/stable/cli/pip_lock/ But who cares about pip, uv is here.
758.
▲
by
esafak
6mo ago
I would leave.
759.
▲
by
esafak
6mo ago
Jujutsu + stacked PRs.
760.
▲
by
esafak
6mo ago
That's why architecture matters.
761.
▲
by
esafak
6mo ago
They're just working at companies with mature products where people are in meetings all day -- they say so! Startups very much want to crank shit out faster.
762.
▲
by
esafak
6mo ago
VCS agnosticism would be a selling point.
763.
▲
by
esafak
6mo ago
How well does it hold up under load? What are the CI and PR stories?
764.
▲
by
esafak
6mo ago
Classic prisoner's dilemma. That's what government is for.
765.
▲
by
esafak
6mo ago
One practice I've heard is reviewing formal work specs like https://specs.md/ or https://github.github.com/spec-kit/ INSTEAD of code. I see value in using a standardized spec throughout a company,
766.
▲
by
esafak
6mo ago
You tried v4?
767.
▲
by
esafak
6mo ago
AI allows you to do things you could not do before so it is fair to say they can't do the new job without AI.
768.
▲
by
esafak
6mo ago
What a bizarre article. He laments the use of AI and then hopes that it might cause a flood of programmers.
769.
▲
by
esafak
6mo ago
What's your easy way of reverse engineering every previous version of this file format, if you don't think that was the hard part?
770.
▲
by
esafak
6mo ago
Facebook's whole conception of tagging is a disaster. It's pure noise. What do I gain by seeing you tag someone (or worse, a bot) in a post? Just share it with them privately.
771.
▲
by
esafak
6mo ago
And you will think less of the people who go there. 30% cheated!!
772.
▲
by
esafak
6mo ago
> It’s sloppy, it’s slow, it’s almost two megabytes in size and somewhat worse than WinRAR on compression. As mathematicians say, optimization is left as an exercise to the reader. You did the hard part.
773.
▲
by
esafak
6mo ago
The title suggests it is a Github issue but really it is https://github.com/composer/composer no? I would edit the title for clarity.
774.
▲
by
esafak
6mo ago
Are you using it to replace Mimir?
775.
▲
by
esafak
6mo ago
I could easily enjoy a 10x improvement when working on something I'm not familiar with once you factor in the learning time.
776.
▲
by
esafak
6mo ago
ty still ahead? When is pyrefly better?
777.
▲
by
esafak
6mo ago
Roughly how big are the datasets?
778.
▲
AI load breaks GitHub – why not other vendors?
(blog.pragmaticengineer.com)
9 points
by
esafak
6mo ago
|
3 comments
779.
▲
by
esafak
6mo ago
I don't understand. Let's stay my state is whether we are in conformance with repo patterns. Walk me through how you don't/can't hallucinate, given that you need an LLM to determine the state. For state variables th
780.
▲
by
esafak
6mo ago
I just have a smart model write a testable phased plan, have a cheaper model implement them, and yet another model to review each phase. I don't see the value of adding a Rust state engine. Algorithmically verifiable things can be test
More ›