i ported some of the FFI experiments i have been hacking on in my lo runtime over to @nodejs. this should work (with many caveats) on linux x64/arm64 (including Raspberry Pi 3B onwards) and macos/arm64. and yes, it's pretty ffast. 🏃🚀 /1

RT @jarredsumner: yeah unfortunately, URLPattern is slow by design. We’ve been reluctant to support it in Bun because APIs that look nice &…

RT @MarcJBrooker: In this month's ACM Queue, @ankushpd and I write about some of the methods and tools we apply to systems correctness at A…

RT @phoronix: IO_uring Zero-Copy Receive Support Ready For Linux 6.15 Networking +29~41% improvement with IO_uring over epoll with benchma…

RT @awesomekling: I'll be speaking about @ladybirdbrowser at the 13th "State of the Browser" in London on March 29th! The event is hosted…

"Having a slow feedback loop has been a major drain on both our productivity and also our enjoyment when working on the code base"

RT @_rafaelgss: One of Node.js security initiatives since last year is to have a Maintainers Threat Model. This documents the access per gr…

RT @robpalmer2: JSR is a package registry with great technical foundations, and now it has Open Governance 🎉 I'm pleased to see energy, ex…

RT @rough__sea: I wrote up my thoughts on Oracle's latest JavaScript trademark filing

RT @lauriewired: Most CPU instructions map 1-to-1 with their microcode, even on x86. A simple mul (multiplication) operation is predictabl…

RT @MichaelArnaldi: which leads us to the conclusion of this X's immense thread, there's two ways of rewriting TS, one is a fool's errand a…

RT @LowLevelTweets: @dillon_mulroy

oh, and i forgot to mention the "assembler" itself, which is still very janky and unfinished, can run on any JS runtime because it's "just js". zero dependencies. it wouldn't be a huge amount of work to add some more useful assembly instrns to it and do all sorts of craziness. 🤪

@nk @nodejs i think it would be tricky to take advantage of v8 fastcalls using the NAPI api. but defo worth taking a look at.

@nk @nodejs it depends i guess. probably won't make a huge difference unless you are really worried about nanoseconds. but they can add up. also, NAPI is pretty much a standard rn and has a lot more options for async tasks etc. than this has rn. it's just a PoC rn.

@nk @nodejs this *should* be significantly faster than napi in microbenchmarks because it's able to use v8 fastcalls. don't quote me on that though. i haven't done any direct comparisons with NAPI across the different runtimes. will try to do so.

@nk @nodejs the code in this repo also lets you generate bindings like NAPI does, but in a very v8 specific way and not cross-runtime way. so there would be concerns around v8 api's breaking compatibility etc. between node releases etc. if you wanted to load the bindings as shared libraries.

RT @deno_land: JSR is now openly governed — meet its board members and check out its governance charter 👇

@Shreyassanthu77 @bunjavascript @deno_land yep. deno ffi is really fast. the approach they took is what inspired me to do this.