October 28, 2025
So fast, even the benchmarks vanished
Show HN: Apache Fory Rust – 10-20x faster serialization than JSON/Protobuf
Promises 10–20x speed, but the crowd hits 404s, asks “where’s the JS?” and doubts no‑schema
TLDR: Apache Fory Rust promises huge speed and easy data sharing across languages, but early readers hit broken benchmark links and asked about browser support and the “no schema” claim. Hype is high, trust is pending: ship working docs, an npm package, and clear benchmarks to win the crowd.
Apache Fory Rust burst onto the scene bragging about being 10–20x faster than popular data formats like JSON and Protobuf, plus magic tricks like keeping “things that point to each other” intact and working across languages with no setup. Sounds dreamy… until the crowd actually clicked. The top vibe? Receipts, please. Multiple readers slammed into dead links, including the supposed Rust benchmarks page, triggering a wave of side‑eye and snark. One wag joked it’s “so fast the docs couldn’t keep up,” as others begged for a simple proof chart. Credibility meter: wobbling.
Then came the practical questions. JavaScript fans wanted to know: is this only for server nerds, or is there a real browser story? Folks poked at the “cross‑language” claim with no npm package in sight, asking where the web fits in. Another thread grilled the bold “no schema” promise—if there’s no shared blueprint, how do different languages agree on what’s what? And the pedants (we love them) zeroed in on messy numbers like NaN and Infinity. Still, there’s genuine excitement: the circular‑reference demo looked slick, and people love the idea of speed without pain. The mood: intrigued but cautious, hyped but hollering for docs, demos, and a real benchmark link they can actually open. See you in the docs—if they load, anyway.
Key Points
- •Apache Fory Rust is introduced as a cross-language serialization framework built in Rust emphasizing ultra-fast performance.
- •It claims to handle complex data needs automatically, including circular references, trait objects, and schema evolution, without IDL files or code generation.
- •The framework uses a shared binary protocol across languages (e.g., Java, Python, C++, Go) to enable serialize-in-one-language, deserialize-in-another workflows.
- •Examples show automatic preservation of shared reference identity and correct handling of cycles using Rust Rc, RcWeak, and RefCell.
- •The title asserts performance advantages over JSON and Protobuf, while the body highlights zero-copy techniques and Rust safety as contributors to speed.