)]}'
{
  "commit": "aec1f2aefb15fc874fcf641eddcaac658f2441bb",
  "tree": "49ba0085b391e93b73295acebe17e9749460adf3",
  "parents": [
    "3c00c96d3af4d5b5e101e56cc161a608b21366ee"
  ],
  "author": {
    "name": "Waffle Lapkin",
    "email": "waffle.lapkin@gmail.com",
    "time": "Mon Jul 06 20:51:32 2026 +0200"
  },
  "committer": {
    "name": "Waffle Lapkin",
    "email": "waffle.lapkin@gmail.com",
    "time": "Tue Jul 07 18:31:33 2026 +0200"
  },
  "message": "simplify `slice::Iter[Mut]::next_chunk` implementation\n\nI verified that this doesn\u0027t pessimize codegen using the example from the\nPR that inroduced the optimization of `next_chunk` (# 149131):\n\n```rust\n#![feature(iter_next_chunk)]\n\n#[no_mangle]\npub fn simd_sum_slow(arr: \u0026[u32]) -\u003e u32 {\n    const STEP_SIZE: usize \u003d 16;\n\n    let mut result \u003d [0; STEP_SIZE];\n\n    let mut iter \u003d arr.iter();\n\n    while let Ok(c) \u003d iter.next_chunk::\u003cSTEP_SIZE\u003e() {\n        for (\u0026n, r) in c.iter().zip(result.iter_mut()) {\n            *r +\u003d n;\n        }\n    }\n\n    result.iter().sum()\n}\n```\n\nI compiled this example with\n```shell\n./build/host/stage1/bin/rustc t.rs --emit\u003dasm -O --crate-type\u003dlib\n```\n\nBefore and after this change; the only difference is the choice of the\njump instruction, which I think shouldn\u0027t make any difference:\n\n```diff\n28,29c28,29\n\u003c \tcmpq\t$64, %rsi\n\u003c \tjae\t.LBB0_2\n---\n\u003e \tcmpq\t$60, %rsi\n\u003e \tja\t.LBB0_2\n```\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "1c8a4d5ba3c2b368dcac98e75716da0ad0a36548",
      "old_mode": 33188,
      "old_path": "library/core/src/slice/iter/macros.rs",
      "new_id": "6f0f7aa7404fa4873a91c26e87fd582dfd6e6bc7",
      "new_mode": 33188,
      "new_path": "library/core/src/slice/iter/macros.rs"
    }
  ]
}