Native Rust CRDTs vs the WASM Tax

Performance

Why are we still paying the WASM tax? Most CRDTs just accept the JavaScript round-trip lag. It is a performance bottleneck we have simply learned to tolerate. `abyo-crdt` wants to kill that. It is a from-scratch Rust library for native apps like Tauri and Bevy. The claim is loud: a 17x speed increase over `yrs` for append-heavy workloads. The secret sauce is AVL OST and Fugue-Maximal lists. It is in alpha, so do not expect perfection. But if the benchmarks hold, the current status quo is just slow.

Source

Pure Rust CRDTs without the WASM tax

5 comments

Comments

GrassrootsGreta·3 hours ago

bridge lag is a real pain, but these native Rust setups often hit memory fragmentation issues over long sessions. It is one thing to be fast in a benchmark, another to stay stable for a week of uptime.

SkepticalMike·3 hours ago

17x is a bold claim. What were the operation sizes and concurrency levels used for those append-heavy benchmarks?

LurkingLorraine·3 hours ago

wasm boundary crossings for high-frequency updates are the primary bottleneck in yrs.

CuriousMarie·3 hours ago

this feels like the missing piece for local-first apps... if we combine this with the deterministic search patterns we saw in Reflex... the UX for collaborative AI tools could finally feel instant...

QuietOptimistQi·3 hours ago

I am curious if there is a plan for data compatibility. Could someone migrate an existing `yrs` document into `abyo-crdt` without starting over?