Zig’s Io.Threaded is neat
matklad.github.io
matklad.github.io
You can interrupt a blocking IO operation with either interrupt() or close().
https://docs.oracle.com/en/java/javase/25/docs/api/java.base...
> In Java, there’s a similarly looking thread interruption mechanism. Critically, it doesn’t support interrupting syscalls: IOException and InterruptedException are both checked and unrelated, meaning that IOing functions are not interruptible
One says IOing functions are not interruptible, other says "You can interrupt a blocking IO operation" :)
https://docs.oracle.com/en/java/javase/25/docs/api/java.base...
The article just references Thread.interrupt
There is a significant caveat to this, which is that if the underlying blocking IO operation isn't natively cancellable (and the JDK doesn't use the trick Zig uses here), then the interrupt is allowed to close the IO resource as a way of ending the operation early. In practice this is usually fine, as you're interrupting the thread precisely because you want it to give up on whatever it's doing, but it precludes some particularly subtle possible IO designs.
Not sure if this changes under virtual threads, where the thread is much more loosely coupled to the syscall.
It actually kind of annoys me; I feel like most Java code I've had to debug/maintain is slow and terrible because most Java engineers lack any ambition so they write the old terrible IO that they learned in university, and for years I guess I kind of erroneously assumed that Java was "bad" at doing IO.
Then I read a book on NIO, started writing my own code with channels and selectors and realized that it actually works quite well, and I found it relatively intuitive and performance was pretty solid.
This is actually a recurring pattern I've found in the Java world; Java has added a lot of great features over the years that make the language more ergonomic and performant and fun to write, but due to its prevalence in the "Enterprise" world and university, it selection biases towards people who refuse to learn anything new, meaning that a lot (the majority?) of Java code feels like it was written in 1997.
I have been writing a lot of Java in my free time, unapologetically using Java 17, 21, and 25 features, and I am actually enjoying it.
Great that Zig is supporting this more “first class”; not strictly required, but looks useful.
There's this line:
> Wouldn’t it be cool if we could just use standard OS threads, blocking APIs, avoid new shinies like io_uring, but still get to cancel any work reliably?
Wish more was said about this: why would that be cool? I get that having robust cancellation in standard threads is cool, but why do we want to avoid io_uring?
io_uring has been one of the greatest sources of exploits for the Linux kernel in years. io_uring has resulted in so many practical privilege escalation attacks that many containers completely block it, and companies like Google keep it entirely disabled. RHEL also ships a kernel without it by default.
This is an example why Go is such a successful language, in my view. It hides the async complexity and allows you to pretend you have a simple continuous thread of execution. It's just cheaper than system threads. Imagine if operating system threads and the blocking syscalls were this efficient.
I've spent the last year building a similar runtime for Zig. Purely because I want my applications to forget they are using things like io_uring in the background.
Disclaimer: I haven’t looked at Zig’s implementation; I’m only going off how the Unix APIs work.
The borrow checker will catch all kinds of bugs that AI-generated code will happily compile in Zig but will blow up as UB at runtime.
Zig's value prop is different and closer to a modern C: it fits in your head and maps fairly closely to assembly. There is no hidden control flow or hidden allocations, so you can tweak performance at a very low level. You're the pilot, not the compiler.
Rust trades the low-level clarity for compiler-enforced safety. If you're not a code artisan or don't care about being one, then please use Rust.
This to me reads like an LLMism.
This has not been my experience; Claude does a great job of writing unit tests for the Zig code, and combined with Zig's fuzzing features and a Python-based integration testing suite, I have avoided hitting runtime UB so far. In exchange, I get much faster compile times, which are important for the agentic development loop.
> Rust trades the low-level clarity for compiler-enforced safety. If you're not a code artisan or don't care about being one, then please use Rust.
Not really seeing the connection here. What do you mean by code artisan? Are you trying to gatekeep Zig because you think Zig is too difficult for LLMs?
Being a code artisan is not gatekeeping. It means if you're used to being close to the metal and sometimes know better than the compiler, Zig is for you. If you're not, use Rust.
With all due respect, that tickles my spidey senses. People claiming to know better than a compiler about low-level details of their code are most often either engaging in premature hyper-optimisation, or suffer a specific kind of greybeard hubris IMHO
A generalised algorithm can do this consistently, billions of times, which happens to be the amount of problems involved in building an application.
The overlap of situations where compiler-second-guessing truly being necessary and the programmers present are capable enough to implement it is narrow enough that I’m fine with most languages’ escape hatches being similarly narrow.
Of course I’d rather be writing the engine. =)
You are saying that if you are not a code artisan (who by definition can't use LLM) likeyou then they should not use Zig. Honestly, I don't care. I'm not a code artisan, if Zig can help me ship faster and better with LLMs, my job is done and that's what I get paid for.
> Codex CLI taking over 15 minutes to recompile on my fairly high-end machine. Zig's incremental compilation optimizations are amazing, giving near-instant recompiles measured in ms instead of minutes.
So there is a benefit here mentioned by the non-arisans which is what I wanted to know but you wouldn't know if you don't use ai assisted delivery.
Comparisons to Rust usually stop at memory safety and overlook what no-hidden-control-flow buys you when it's applied consistently across the whole language.
The same mechanism that lets you control allocation (passing an allocator) extends to I/O. For instance, writing Zig, I know exactly when I'm handing control to the kernel and when I'm not. In the vast majority of languages, it's something you'd observe at runtime, whereas in Zig, it's simply the source code you compiled.
So the point of writing Zig over any other language isn't beating the compiler, it's understanding and controlling what your program does at any given moment.
In general, what helps is to instruct ai to use test coverage to ensure the cover all edge cases with tests.
> Zig's value prop is different and closer to a modern C: it fits in your head and it maps fairly closely to assembly.
This can't be correct for the simple reason that it has both a modern optimizing compiler (LLVM) and UB.
A modern compiler will try to autovectorize your code, so you don't really know what the assembly will be.
UB gives compiler license to rewrite your code however it sees fit.
Looking at what the compiler actually did is an ancient tradition. I’ve even done it with Java. Seeing five virtual calls with bounds checking all over get turned into a single load instruction has to be seen to be believed.
Like programming in a kind of super strict IL. Or the opposite : super poweful, super abstract language, yet extremely strict.
cargo clean
cargo check
Finished `dev` profile [unoptimized + debuginfo] target(s) in 2m 40s
cargo build
Finished `dev` profile [unoptimized + debuginfo] target(s) in 3m 12s
cargo clean
cargo build
Finished `dev` profile [unoptimized + debuginfo] target(s) in 3m 15s
Taking the alacritty project and modifying the source files to pretend something happened:
> cargo clean
Removed 1602 files, 545.8MiB total
> cargo check
[...]
Finished `dev` profile [unoptimized + debuginfo] target(s) in 17.12s
> cargo build
[...]
Finished `dev` profile [unoptimized + debuginfo] target(s) in 23.04s
> find alacritty_terminal/ -name "*.rs" | xargs-I{} sh -c 'echo >> "{}"'
> cargo check
[...]
Finished `dev` profile [unoptimized + debuginfo] target(s) in 1.00s
> cargo build
[...]
Finished `dev` profile [unoptimized + debuginfo] target(s) in 2.37s
The question I ask myself is “would I write a game in Rust?” Zig seems like a better fit: you can be clever without being chided by the borrow checker. But as a “code artisan”, there’s an interesting challenge in making it work with Rust.
I might use Zig on a project if it was just me. I wouldn’t use it with the a team that I didn’t hand pick.
Smart people like Zig. Smart people like Rust. Smart people like LLMs. It’s a big Venn diagram. It be great if we could not bash each other.
Are you able to ship faster with Zig vs Rust? How did you arrive at the 4x figure?
I'm really puzzled why people are so angry over using ai assistance with Zig. LLMs are always going to be able to write better code faster and better than anyone who calls themselves code artisans in this day and age.
I think this comment posted a while back someone who worked on Zig is pretty telling:
> The key problem with Zig nowadays is how much of its community and adoption is driven by anti-Rust sentiment. As a result, while Rust puts beginner onboarding and documentation at the center of its culture, as opposed to the “C neckbeard”'s culture, Zig is going the other way around.
And ultimately this didn't stop people from using AI generating Zig code as much of the users in the thread are already using it to ship quality faster.
Happy to use the work the community outputs and culture is largely irrelevant when prompting LLMs outputs code that doesn't really require any intimate knowledge with the syntaxes
LLMs can't write better code faster than any reasonably competent human. They suck at writing code, and if you're outsourcing programming to them the software you produce will suck as well.
Do we even care about compilation time if we set it to work on a task and return some time later to see the result?
---
Edit for replies complaining I used debug.print:
Okay, `try std.io.getStdOut().writer().print("Hello, world!\n", .{});`. Still one line, more explicit with chained methods, but that still just goes to show you how arbitrary this metric is -- Zig's standard library makes you be more explicit with one function, but it doesn't make you be as explicit with another one that works essentially the same way but writes to a different stream, ergo this has nothing to do with the language's inherent explicitness but rather comparing the implementation of a single random function.
In Zig, you create a buffered writer for the stdout file handle, write hello world to the buffer, and then flush it. It exposes what's actually happening underneath.
small typo fix:
instead of `std.io`, it should be `init.io`. `init` is first parameter in `main` function.
When I had problems in my Zig codebase(latest version), the LLM cant resolve my problem. It kept giving me results based on older versions of the Zig code. This was due to breaking changes in every version. Thats why I no longer ask the LLM for help with Zig-related code. The folks on Zig Discord server and its forum, helps me very much.
Sure it wastes a bit of time always figure it out when it fails to compile.
This is why I’m excited to see qwen3.8 27b removed a bunch of knowledge for more reasoning. Baked in old zig api is worthless
Zig has benefits, like comptime, no hidden control flow, memory mngmt, cross compilation etc. But its syntax is just a mouthfull of wats, whys and wtfs.
I can forsee a compile to Zig languge popping up sometime soon.
in general, I like zig syntax, but i find that it had very little regard to how the eye moves on the page. it’s jarring at times.
if (evaluation) result = evaluation
if (evaluation) |result|
Although assignments as expressions in C are convenient, I don't think they make a ton of sense and find Zig's capture easier to read (easier to immediately see what's being evaluated)
resultA : resultB ? (evaluation)
if we always put result first?
I think the most important piece of if statements and ternaries is the condition being evaluated and it's worth putting that first?
and ternary is a branch with no variable introduction, so it doesn’t apply again.
(it’s basically a weak switch expression, and the value moves leftwards towards its invocation.)
To ship what I do now would've cost a few million dollars per year in terms of the team needed to maintain it now its 1000x cheaper
My argument was that syntax still matters because LLMs communicate in syntax, and they pay for that per token, therefore LLMs make syntax more relevent as you can actually calculate cost to write in one language versus another.
I'm not arguing that humans can compete with a network of agents.
If you don't care, don't go around in programming threads telling you how it doesn't matter anymore. It doesn't matter to you, because you prefer to be ignorant about the details of programming, and you have found tools that let you live in ignorance, good for you. But some of us do care, and are going to continue to do so.
LLMs compile english to whatever code you want and it can write it better, faster and more efficiently than a human can over the long run. You might have an edge in shorter sprints but that gap is quickly reduced when LLMs simply learn and apply whatever optimization they find and it can continue to output without any diminishing returns and at fraction of what it takes to pay a human to do it.
But that's just my opinion.
Such questions usually come down to "how familiar am I with this programming language", and anything that doesn't look like your pet language looks ugly. Except Rust and C++ of course, those are objectively ugly ;P
This I don't understand? Zig structs are pretty much the same as in any other language, except that the declaration can be more flexible by building the struct type via comptime code (e.g. how Zig does generics) - and Zig groups structs and namespaces under the same keyword (which hardly matters in practice though).
> @SpecialThing
...not any weirder than `__builtin` keywords in Clang or GCC and at least in Zig it's obvious that anything starting with a `@` is a builtin, while in Clang/GCC the `__` is not explicitly reserved for builtins. ObjC also uses a leading `@` to separate ObjC keywords from regular C keywords. Agreed that the @-noise can get a bit excessive in Zig though (mainly because implicit casting is heavily restricted, at least compared to C).
> Multiline strings are also a weird addition
AFAIK the 'weird' multiline syntax is for keeping the parser simple and fast. Multi-line-strings themselves are a good feature to have though.
In general, if Zig code starts to look too 'noisy', that's usually a sign that the code should be simplified.
I definitely don't agree with all design decisions in Zig, but the syntax is mostly fine (my main critique points are that Zig often tries to lean too much into 'design purity' than 'programmer convenience' and that parts of the stdlib are too 'object oriented'.
In fact in C identifiers starting with _ are reserved for the implementation (except for in local scope if the second character is not an underscore or capital letter).
~ and @ weren't replaced by anything with ::, though, ~Foo is Box<Foo> and @Foo is Rc<Foo> (or is it Arc<Foo>?). :: to access a member of a module already existed when ~ and @ was in Rust.