Jonathan S gereeter SHA512: ae0fa3af71c32c4885b8b0dead0e949f3b3864f463e581266c9788fa1b8f1c87fde7187b78fc5deedf5748d3e6852d6a3f113d417d80e06838c8d381d57afa6a'

Gankra/collect-rs 65

Miscellaneous Collections

gereeter/bounded-intmap 18

A reimplementation of `Data.IntMap` that uses minimum and maximum bounds on subtrees instread of bit prefixes.

gereeter/csvsheet 2

A terminal-based editor for CSV files

gereeter/bad-crypto 1

Playing around with crypto in Rust

gereeter/btree-rewrite 1

A rewrite of Rust's BTreeMap

gereeter/chunked-conduit 1

Perfect streaming in conduit using chunking

gereeter/ajhc 0

A fork of jhc.

gereeter/arrayvec 0

A vector with a fixed capacity. (Rust)

gereeter/capstone 0

Capstone disassembly/disassembler framework: Core (Arm, Arm64, M68K, Mips, PPC, Sparc, SystemZ, X86, X86_64, XCore) + bindings (Python, Java, Ocaml)

issue commentziglang/zig

rename FloatMode.Optimized to FloatMode.Mathematical

This seems misleading; I'd expect a "mathematical" mode to act predictably and follow mathematical laws, or maybe improve numerical accuracy, but that's not what the fast math flags do. They instead break more laws (technically you can't even rely on x == x to be true, for example) and can sometimes worsen numerical accuracy in search of performance.

Perhaps this would make sense if @setFloatMode was renamed to @setFloatAssumptions?


comment created time in 11 days

Pull request review commentziglang/zig

std.rand: set DefaultCsprng to Gimli, and require a larger seed

 pub const Gimli = struct {     random: Random,     state: std.crypto.core.Gimli, -    pub fn init(init_s: u64) Gimli {+    pub const secret_seed_length = 32;
    pub const secret_seed_length = std.crypto.core.Gimli.BLOCKBYTES - std.crypto.core.Gimli.RATE;

would make it more clear that the seed is exactly as long as the preserved state between calls.


comment created time in 11 days