March 18, 2026
JSON wars ignite again
RX – a new random-access JSON alternative
Promises tiny data and lightning speed; commenters clap, scoff, and shout “not drop‑in”
TLDR: RX promises much smaller data and ultra-fast lookups without loading everything, aiming to replace everyday JSON. The crowd split fast: fans cheer the speed, skeptics say it isn’t truly “drop-in” and question ditching human-readability, turning it into a showdown of performance versus simplicity.
A new challenger just slid into the JSON DMs: RX, a “random‑access JSON alternative” from the creator of nvm, claims up to 18x smaller data and “23,000x faster” lookups by peeking directly into the bytes. The demo is slick, the CLI is cute, and the GitHub readme glows. But the internet? Oh, it had thoughts.
First punch came from the performance purists: “You shouldn’t be using JSON for performance,” snapped one commenter, arguing this solves the wrong problem. Then the readability defenders piled in. If JSON is beloved for being human‑readable, why compare it to a binary format at all? “Is everything a ‘JSON alternative’ now?” groaned another, rolling their eyes at the branding. Meanwhile, optimists cheered the idea of a faster, smaller drop‑in and even compared it to EXI (the “efficient” version of XML) — praying one of these next‑gen formats finally wins.
And then the twist: the “drop‑in” claim sparked a mini‑uproar. Because RX’s decoded result is a read‑only Proxy, a sharp‑eyed dev pointed out that code relying on mutating parsed objects might break. Cue the chorus: is “drop‑in” fair? The vibe: Team Speed vs Team Simple, with a side of “lawyer up those words.” Memes flew about “23,000x faster… for the one key you actually need.” Hot, spicy, chaotic — just how devs like it.
Key Points
- •REXC provides drop-in stringify/parse alternatives to JSON with random access and reduced parsing overhead.
- •Benchmarks on a 35,000-key dataset report up to 18x smaller output and up to 23,000x faster single-key lookup.
- •The parsed result is a read-only Proxy over a flat byte buffer, aiming for near-zero heap allocations.
- •The library offers both string and binary APIs, plus a CLI for conversion, pretty-printing, and sub-value selection.
- •Encoding options include sorted indexes, external reference dictionaries, and streaming output via onChunk.