Draft release notes for 1.87

[BoxyUwU]

NOTE: We are trying this issue as a new way to work on the draft release notes.
Use the 📝 links to edit those that have a relnotes-tracking-issue,
and they will be updated when we regenerate the notes periodically.

cc @rust-lang/release

---

Version 1.87.0 (2025-05-15)

Language

• Stabilize asm_goto feature
  📝
• Allow parsing open beginning ranges after unary operators
  📝
• Don't require method impls for methods with Self: Sized bounds in impls for unsized types
  📝
• Stabilize feature(precise_capturing_in_traits) allowing use<...> bounds on return position impl Trait in traits
  📝

Compiler

• x86: make SSE2 required for i686 targets and use it to pass SIMD types
  📝

Platform Support

Refer to Rust's platform support page
for more information on Rust's tiered platform support.

Libraries

• Stabilize the anonymous pipe API
  📝
• Add support for unbounded left/right shift operations
  📝
• Print pointer metadata in Debug impl of raw pointers
  📝
• Vec::with_capacity guarantees it allocates with the amount requested, even if Vec::capacity returns a different number.
  📝
• Most std::arch intrinsics which don't take pointer arguments can now be called from safe code if the caller has the appropriate target features already enabled (mark x86 intrinsics as safe stdarch#1714, mark arm intrinsics as safe stdarch#1716, mark riscv intrinsics as safe stdarch#1717)
  📝
• Undeprecate env::home_dir
  📝
• Denote ControlFlow as #[must_use]
  📝
• Macros such as assert_eq! and vec! now support const {...} expressions
  📝

Stabilized APIs

• Vec::extract_if
• vec::ExtractIf
• LinkedList::extract_if
• linked_list::ExtractIf
  📝
• <[T]>::split_off
• <[T]>::split_off_mut
• <[T]>::split_off_first
• <[T]>::split_off_first_mut
• <[T]>::split_off_last
• <[T]>::split_off_last_mut
  📝
• String::extend_from_within
  📝
• os_str::Display
• OsString::display
• OsStr::display
  📝
• io::pipe
• io::PipeReader
• io::PipeWriter
• impl From<PipeReader> for OwnedHandle
• impl From<PipeWriter> for OwnedHandle
• impl From<PipeReader> for Stdio
• impl From<PipeWriter> for Stdio
• impl From<PipeReader> for OwnedFd
• impl From<PipeWriter> for OwnedFd
  📝
• Box<MaybeUninit<T>>::write
  📝
• impl TryFrom<Vec<u8>> for String
  📝

These APIs are now stable in const contexts:

• <*const T>::offset_from_unsigned
• <*const T>::byte_offset_from_unsigned
• <*mut T>::offset_from_unsigned
• <*mut T>::byte_offset_from_unsigned
• NonNull::offset_from_unsigned
• NonNull::byte_offset_from_unsigned
  📝
• <uN>::cast_signed
• NonZero::<uN>::cast_signed.
• <iN>::cast_signed.
• NonZero::<iN>::cast_unsigned.
  📝
• <uN>::is_multiple_of
  📝
• <uN>::unbounded_shl
• <uN>::unbounded_shr
• <iN>::unbounded_shl
• <iN>::unbounded_shr
  📝
• <str>::from_utf8
• <str>::from_utf8_mut
• <str>::from_utf8_unchecked
• <str>::from_utf8_unchecked_mut
  📝
• <[T]>::copy_from_slice
  📝
• SocketAddr::set_ip
• SocketAddr::set_port,
• SocketAddrV4::set_ip
• SocketAddrV4::set_port,
• SocketAddrV6::set_ip
• SocketAddrV6::set_port
• SocketAddrV6::set_flowinfo
• SocketAddrV6::set_scope_id
  📝
• char::is_digit
• char::is_whitespace
  📝
• <iN>:midpoint
  📝
• <[[T; N]]>::as_flattened
• <[[T; N]]>::as_flattened_mut
  📝
• <str>::from_utf8_mut
• core::str::from_utf8_mut
  📝
• String::into_bytes
• String::as_str
• String::capacity
• String::as_bytes
• String::len
• String::is_empty
• String::as_mut_str
• String::as_mut_vec
• Vec::as_ptr
• Vec::as_slice
• Vec::capacity
• Vec::len
• Vec::is_empty
• Vec::as_mut_slice
• Vec::as_mut_ptr
  📝

Cargo

• Add terminal integration via ANSI OSC 9;4 sequences
• chore: bump openssl to v3
• feat(package): add --exclude-lockfile flag

Rustdoc

Compatibility Notes

• Rust now raises an error for macro invocations inside the #![crate_name] attribute
  📝
• Unstable fields are now always considered to be inhabited
  📝
• Macros ??? may ?? break now ??? due to ranges and unary operators ??? idk ?
  📝
• Make Debug impl of raw pointers print metadata if present
  📝
• Warn against function pointers using unsupported ABI strings in dependencies
  📝
• Associated types on dyn types are no longer deduplicated
  📝
• Forbid attributes on .. inside of struct patterns (let Struct { #[attribute] .. }) =
  📝
• Make ptr_cast_add_auto_to_object lint into hard error
  📝
• Many std::arch intrinsics are now safe to call in some contexts, there may now be new unused_unsafe warnings in existing codebases.
  📝
• Limit width and precision formatting options to 16 bits on all targets
  📝
• Turn order dependent trait objects future incompat warning into a hard error
  📝
• Denote ControlFlow as #[must_use]
  📝
• Remove i586-pc-windows-msvc target
  📝
• Windows: The standard library no longer links advapi32, except on win7. Code such as C libraries that were relying on this assumption may need to explicitly link advapi32.
  📝
• Remove Nonterminal and TokenKind::Interpolated
  📝
• expand: Leave traces when expanding cfg attributes
  📝

Internal Changes

These changes do not affect any public interfaces of Rust, but they represent
significant improvements to the performance or internals of rustc and related
tools.

• Update to LLVM 20
  📝

Other

[BoxyUwU]

@rustbot ping relnotes-interest-group

[rustbot]

Hi relnotes-interest-group, this issue/PR could use some help in reviewing /
adjusting release notes. Could you take a look if available? Thanks <3

cc @alex-semenyuk @jieyouxu @joshtriplett @Kobzol @lcnr @traviscross

[elizabethdwyer255]

The link to

• <str>::from_utf8_unchecked

under "These APIs are now stable in const contexts:" needs to be fixed. Currently it's written as

• [<str>::from_utf8_unchecked](https://doc.rust-lang.org/nightly/std/primitive.str.html#method.from_utf8_unchecked]

i.e. the final parenthesis is accidentally written as a square bracket.

[miestrode]

Is there a reason why <iN>:midpoint is written with only a single colon? Never commented here before so I might be wrong, but this seems like a mistake to me.

[jieyouxu]

Is there a reason why <iN>:midpoint is written with only a single colon? Never commented here before so I might be wrong, but this seems like a mistake to me.

It's a typo, I fixed the relnotes (it will be correct the next time the listing gets sync'd), thanks