systemd Opens the Door to Rust

With security vulnerabilities being reported at an ever-growing pace, moving more and more of Linux to Rust is a good path to be on. systemd sits at the core of nearly every Linux system, and on 24 October Christian Brauner merged support for writing systemd programs in Rust. After the kernel and Ubuntu's Rust coreutils and sudo (see From Lab to Factory), it is the next big piece of Linux to let Rust in.

Three days later the first tool landed. systemd-keyring-setup enrolls X.509 certificates into the kernel's keyrings at boot and then seals them. The maintainers set tight limits:

• Rust is only for new tools at the edge of systemd, and existing C code stays
• There is no cargo and there are no outside crates, everything builds with meson as no_std code
• systemd brings its own Box and Vec, where a failed allocation returns an error instead of aborting

I wrote up the details in my full post on rust-trends.com.

Much of the work on the Rust compiler happens out of sight, which is exactly why I enjoyed Matyáš Racek’s post. Compiler performance is his main interest, he recently joined perf triage, and his point is that several long-running efforts pay off in the same year.

The next-generation trait solver, almost ten years in the making, fixes a batch of soundness bugs and holds answers to old problems in async Rust and impl Trait. Polonius, the new borrow checker, lets more code compile. The parallel front end is close to stabilization. Racek calls LLVM 23 an "unambiguous and unexpectedly large" performance win for rustc. A project goal also aims to let cargo share build artifacts between cargo check and cargo build.

These surveys shape the language, so it is important they get the reach they need. The Variadic Generics Micro Survey drew over 3,300 submissions. The team read the free answers by hand, for privacy and quality reasons, which is why the results took a year.

One need comes up again and again: implementing a trait for tuples of every size. Today that takes a macro that stops at some fixed length. Skeptics fear "death by a thousand cuts", small obscure features piling up until Rust is hard for newcomers. The team's advice for a future proposal is a syntax that is easy to read and clear wins over the macro workarounds.

TypeScript has earned its place on the web, and with Rust powering the plumbing underneath, it only gets faster. Cloudflare and VoidZero are building tsrs, a Rust port of the TypeScript compiler, in close collaboration with the TypeScript team. It lives in the Oxc toolchain and follows TypeScript's design decisions. Nobody is forking the language.

Early benchmarks put it at up to 8x faster than TypeScript 7.0 with about 70% less memory, on the same number of type-check workers. Why Rust when a Go port already exists? WebAssembly. A Go program has to carry its garbage collector into Wasm and a Rust program does not, which counts when you want a whole toolchain running in a browser.

• Google's Rust Style Guide How Google wants Rust written across Android, Chrome and the rest. It builds on rustfmt and the Rust API Guidelines, then adds firm picks of its own, such as preferring unique ownership over Rc and Arc.

• A fast and secure tar parser for the Rust ecosystem Astral built tar-codec after finding that uv and pip could see different files in the same tarball. Since uv 0.13 it parses source distributions, roughly twice as fast as the parser it replaced.

• How to fund Rust contributors The Rust Funding team lists three ways to pay for Rust work: the maintainer funds of the Rust Foundation and RustNL, fundable Project Goals, and GitHub Sponsors for individual contributors.