Anonymous
8/22/2025, 8:20:58 PM
No.106349298
>>106348397
That case is interesting because the bug was for a type like this where u8 has align=1.
#[repr(align(2))]
struct Foo(u8);
After a quick search, there's not much use of read_unaligned like that anymore in std/core outside of some Windows file stuff and ASCII/UTF8 which shouldn't run into issues like that.
There are still a lot of unsafe things that they're deciding on the rules for, so some code that used to be safe might become UB when they decide on the documentation for it.
https://github.com/rust-lang/unsafe-code-guidelines
>>106349076
>please ensure that "unsafe" is more isolated, or needed less often and not needed for algorithms, or easier to reason about and learn and not unspecified.
Solve the halting problem while you're at it.
That case is interesting because the bug was for a type like this where u8 has align=1.
#[repr(align(2))]
struct Foo(u8);
After a quick search, there's not much use of read_unaligned like that anymore in std/core outside of some Windows file stuff and ASCII/UTF8 which shouldn't run into issues like that.
There are still a lot of unsafe things that they're deciding on the rules for, so some code that used to be safe might become UB when they decide on the documentation for it.
https://github.com/rust-lang/unsafe-code-guidelines
>>106349076
>please ensure that "unsafe" is more isolated, or needed less often and not needed for algorithms, or easier to reason about and learn and not unspecified.
Solve the halting problem while you're at it.