@WG-prioritization/alerts issue #80925 has been requested for prioritization.
This sounds very specific
But indeed it's unsound
I'd say this is not
P-critical because it's so specific (I mean, who uses DSTs and
#[repr(packed)] in the same struct?)
agree with p-medium for this
Issue reporter opened a list of issues related to how Rust represents data layout (at least my limited understanding).
I probably don't have the right context , just one question - is it ok if different issues in this set are assigned different priority levels? At a cursory look I see there are unsound behaviours when Rust is instructed to represent differently data layout and apparently this also varies depending on which compiler I am on
I think so, yes - I kind of disagree that https://github.com/rust-lang/rust/issues/81996 is unsound, it could only cause unsoundness if someone using
unsafe makes assumptions about the layout that to my knowledge aren't guarenteed by the compiler. In this instance, it is most certainly unsound because you don't need unsafe.
ah interesting, thank you Joshua!!
this stuff is a bit above my head so It's great to hear and compare opinions
left a comment on #81996 asking the author to explain the unsound label
Issue #80925's prioritization request has been removed.
Although I removed the
A-const-eval label, I'm uncertain if that was correct now... The second piece of code that @mahkoh pointed out, according to https://doc.rust-lang.org/nightly/nightly-rustc/rustc_mir/interpret/index.html, is really const-eval related.