| website | https://blog.sunfishcode.online |
| github | https://github.com/sunfishcode |
Dan Gohman
- 5 Followers
- 140 Following
- 85 Posts
Toots mostly about WebAssembly, WASI, Wasmtime, Rust, Capability-based security. Fastly. Ex-Mozilla.
@ttraenkler Check out `wasm-tools compose` in https://github.com/bytecodealliance/wasm-tools/
It's still in development and there's not a lot of documentation yet, or inlining. That said, I expect interface types will be advantageously inlinable, because the lifting and lowering encodings give compilers a high-level view of what's going on.
GitHub - bytecodealliance/wasm-tools: CLI and Rust libraries for low-level manipulation of WebAssembly modules
CLI and Rust libraries for low-level manipulation of WebAssembly modules - GitHub - bytecodealliance/wasm-tools: CLI and Rust libraries for low-level manipulation of WebAssembly modules
@repi It's been pretty stable. 1.0 marks the switch to use the OwnedFd etc. from std, but users shouldn't see any difference.
Looking forward to regular open office hours today, but there's a time change: due to daylight savings time here, the time in UTC has shifted an hour later.
cap-std is now at 1.0!
The main change in this release is the update to io-lifetimes 1.0, which marks the update to use the I/O safety features in the standard library when available.
io-lifetimes is now at 1.0!
The main functionality of OwnedFd/BorrowedFd/AsFd/etc. is stable in std as of Rust 1.64. io-lifetimes 1.0 just re-exports those, when available, and provides its own implementation on older Rust versions.
@dherman I don't know of any work on this, but I'm curious: is your intuition that the training set of statically-typed code is more regular, or that the model could be guided or somehow factored by the type system?
Tired: Unicode is complex, let's use ASCII
Wired: Unicode is complex, let's pass around handles to resources rather than passing around names
Wired: Unicode is complex, let's pass around handles to resources rather than passing around names
@yosh dunno if this is the issue, but one thing i've run into doing something similar is that mastodon.social is sometimes really slow right now, so things which involve lots of queries can be reaaally slow
My presentation slides from the Wasm SIMD Subgroup meeting, proposing an approach for permitting nondeterministic SIMD operations, but preserving a path to deterministic semantics: