Add stronger alternatives to `align_to` #105296

oli-obk · 2022-12-05T10:36:57Z

r? @ghost

some code for the discussion

ACP: rust-lang/libs-team#164

library/core/src/slice/mod.rs

Lokathor · 2022-12-05T13:58:04Z

One fn seems to change to bigger elements and the other to smaller elements. I'm not clear why it's two separate functions, and also the naming doesn't really suggest which is which.

scottmcm · 2022-12-05T18:58:54Z

Minor naming thought: Would it be worth leaving space for a future safe-transmute version to have the "ideal" name, and have the unsafe versions be _unchecked or similar?

(Perhaps transmute is enough to communicate the unsafety, though, and we'll use a different word for the safe versions.)

oli-obk · 2022-12-21T11:43:18Z

One fn seems to change to bigger elements and the other to smaller elements. I'm not clear why it's two separate functions, and also the naming doesn't really suggest which is which.

the transmute* methods are now for all kinds of size changes and document where they won't work (if the alignment gets bigger) and suggest to use the other method instead.

library/core/src/slice/mod.rs

jswrenn · 2022-12-21T20:23:33Z

Minor naming thought: Would it be worth leaving space for a future safe-transmute version to have the "ideal" name, and have the unsafe versions be _unchecked or similar?

(Perhaps transmute is enough to communicate the unsafety, though, and we'll use a different word for the safe versions.)

FWIW, I'm leaning towards suggesting that the safe transmute APIs use the word "reinterpret" instead of "transmute".

scottmcm · 2022-12-21T20:57:46Z

@jswrenn Sounds great -- I've long been a fan of that choice :)

rustbot · 2023-01-17T08:57:43Z

Hey! It looks like you've submitted a new PR for the library teams!

If this PR contains changes to any rust-lang/rust public library APIs then please comment with @rustbot label +T-libs-api -T-libs to tag it appropriately. If this PR contains changes to any unstable APIs please edit the PR description to add a link to the relevant API Change Proposal or create one if you haven't already. If you're unsure where your change falls no worries, just leave it as is and the reviewer will take a look and make a decision to forward on if necessary.

Examples of T-libs-api changes:

Stabilizing library features
Introducing insta-stable changes such as new implementations of existing stable traits on existing stable types
Introducing new or changing existing unstable library APIs (excluding permanently unstable features / features without a tracking issue)
Changing public documentation in ways that create new stability guarantees
Changing observable runtime behavior of library APIs

oli-obk · 2023-01-17T08:58:08Z

@rustbot label +T-libs-api -T-libs

oli-obk · 2023-01-17T09:08:20Z

r? libs-api

the8472 · 2023-01-17T12:08:48Z

library/core/src/slice/mod.rs

+    #[must_use]
+    #[unstable(feature = "slice_align_to_ish", issue = "none")]
+    #[inline]
+    pub fn aligned_subslice(&self, align: usize) -> &[T] {


Could the align be const? Does one ever need dynamic alignment?

the8472 · 2023-01-17T12:09:14Z

library/core/src/slice/mod.rs

+        if size_u > size_t {
+            assert!(
+                self.len() * size_u % size_t == 0,
+                "input slice does not fit exactly into a slice of the output element type"
+            );
+        } else {
+            assert!(
+                self.len() * size_t % size_u == 0,
+                "input slice does not fit exactly into a slice of the output element type"
+            );
+        }


Do the panics get eliminated when the types are exact multiples in size and alignment?

no, I should optimize that.

anden3 · 2023-04-05T16:08:39Z

Hello @oli-obk! I just want to ping you as part of the triage procedure to let you know that this PR has received reviews :)

oli-obk · 2023-04-06T13:07:04Z

The PR is blocked on rust-lang/libs-team#164

scottmcm · 2023-11-07T19:37:07Z

library/core/src/slice/mod.rs

+            align % size == 0,
+            "aligned_subslice only works for alignments that are multiples of the element's size"
+        );
+        let offset = self.as_ptr().addr() % align;


If this align is ever a non-constant it'll have to actually emit the division.

Consider taking https://doc.rust-lang.org/std/ptr/struct.Alignment.html instead so you can optimize based on it always being a power of two.

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-libs Relevant to the library team, which will review and decide on the PR/issue. labels Dec 5, 2022

RalfJung reviewed Dec 5, 2022

View reviewed changes