News for hackers. Stuff that matters.

roetlich

Re:My Thoughts on the Bun Rust Rewrite on Friday July 10, 2026 @06:24AM (#48858059)
Love the exclamation marks. So let's just use, for example, this definition: https://www.britannica.com/topic/ad-hominem

Here it's a type of argument, and it's only sometimes a fallacy. Seems like this how it's defined most of the time, unless you explicitly look for "ad hominem fallacy". Ad hominem without any context can be vague, I guess, so I tried to be more explicit.

I don't know, maybe I'm wrong, English isn't my first language, but I think "Obvious Fabrication" (capital letters) is a bit silly.

Re:My Thoughts on the Bun Rust Rewrite on Thursday July 09, 2026 @11:40AM (#48847748)
Oh yes, I didn't clarify that I meant an argumentum ad hominem, which isn't the same thing as the informal fallacy of the same name. So yeah, I agree, this isn't directly fallacious.
Re:My Thoughts on the Bun Rust Rewrite on Thursday July 09, 2026 @09:21AM (#48845402)
> if Andrew isn't trying to enter the discussion

But he is. Both in the title, and in later half of the blog post, he directly enters that discussion. I'm not saying an ad hominem argument is always bad, but this clearly is one.

Re:My Thoughts on the Bun Rust Rewrite on Thursday July 09, 2026 @08:24AM (#48844759)
No, the discussion started with a the article from Bun, stating that rust has some technical advantages for them. The response from the Zig creator is a bunch of personal attacks directed at one guy, like calling him a stinky manager. All of these are fully unrelated to which language is better for Bun.

This is like the textbook definition of an ad hominem.

Re:IBM Debuts First Sub-1 Nanometer Chip Technology on Thursday June 25, 2026 @06:29PM (#48680031)
Well, this seems like a very dangerous way of using AI. If you keep pushing until you get back an answer that makes sense to you, you might just get your own believes fed back to you.
Re:GitHub Copilot App on Tuesday June 02, 2026 @02:38PM (#48374357)
Who would have thought that git worktree is the technology of the year 2026?
Great, now we need this with current data for modern governments
Great idea, to include learning into the regular work.

Can you explain how the lessons are generated?

It looks like there are no pre-generated lessons, they are created on the fly by the AI. Wouldn't it be possible to write a skill or similar that does the same with your existing coding agent?

Re:Graphs That Explain the State of AI in 2026 on Saturday April 18, 2026 @07:30PM (#47820443)
My Firefox doesn't accept your https cert. Maybe check that out?
What does "done offline" mean? Otherwise you are limited by context window.
Re:LibreOffice and the Art of Overreacting on Thursday March 26, 2026 @10:56AM (#47531269)
"Managment" usually are office workers.
Re:OpenAI – How to delete your account on Saturday February 28, 2026 @11:19AM (#47197100)
> Does it continue if you agree to the warning?

No, you can't "agree".

> What if you ask about 9/11?

It answers the question.

Re:Crafting Interpreters on Thursday January 15, 2026 @08:44AM (#46632436)
Well, most interpreters use a compiler internally, for example to compile to byte code. The book explains that as well, so I'd recommend just reading it: https://craftinginterpreters.com/a-map-of-the-territory.html...
Re:Ask HN: Could you share your personal website here? on Wednesday January 14, 2026 @04:51PM (#46624160)
Thanks, will fix! Haven't maintaineed the website in a bit :)
Re:C++ Web Server on my custom hobby OS on Friday November 28, 2025 @12:19PM (#46080628)
Very cool! Why was and the entire networking stack straight forward, but not HTTP (and TCP)? Could you take inspiration form other projects for things like DNS?
Interesting! The article talks mostly about how this all worked, but rarely about what was actually discussed. Which opinions of the party do you like or support?
> one door per day until it's Christmas eve

That would be 24.

Re:Dumb Pipe on Tuesday July 29, 2025 @06:56AM (#44721695)
What?

I think there must be some misunderstanding? I think deathanatos just wants an easy way to send files between computers when the internet is down, which seems decently reasonable.

Re:Brave blocks Microsoft Recall by default on Wednesday July 23, 2025 @08:52AM (#44658659)
Then don't turn it on?
Any comparison to Typst?
Re:Redis is open source again on Friday May 02, 2025 @10:27AM (#43870242)
Unless you modify Redis, what do you want to share? You don't have to share any of your server code or anything. No reason to be so dramatic. :)
Re:Growing a Language [pdf] (1998) on Friday April 18, 2025 @02:17PM (#43730553)
Or you could use C# and get all of those features.
This looks very cool, maybe I'll try this for risc-v if I have some time. I guess you could also use this to write RAW images, or other binary data.
Fair, thanks.
Oh, that makes more sense. Thank you!

Is this linked list type then mostly meant to be used as an internal implementation detail, and not be exposed in a public API? Because if I see a function that takes a non-generic list I won't really know how to call it. :)

Thank you, that answers all my questions.
Well I personally almost never use linked lists, don't like them. But Zig seems to think otherwise, they have two implementations in the standard library.

I don't know much Zig, I just wanted to ask how to use the type that the article talks about?

Let's use the very simple example from the article. Let's say I want to extract this code into a function:

    while (node) |n| {
      const user: *User = @fieldParentPtr("node", n);
      std.debug.print("{any}\n", .{user});
      node = n.next;
    }
1. How does that look, what's the type signature of this function? 2. What happens if I put in a list doesn't contain users? Do I get a simple to understand compile time error, or can I segfault because I'm accessing bad memory?

And I don't think that would need any generics, since the list type isn't generic, right?

Okay, thanks for explaining more.

> And it's typesafe: if you write invalid code the compiler will give you an error.

One more question, if @fieldParentPtr("node", n) is typesafe and can get you a User, the compiler needs to know that the struct has the fields of a User, and that this pointer is indeed pointing at a node field, right? Then why do you need to specify "node" at all?

I think I don't understand Zig at all :)

This looks somewhat horrifying. How do I safely write a function that takes a linked list as a parameter?

Older »