#rust-lang

ubsan: idk, I can't see the problem myself (with ref_slice); I consider ref_tuple questionable except that I think that we ought to guarantee more when it comes to repr, since I suspect in practice we'll wind up committed anyhow, but we ought to make it as an active decision for sure

nmatsakis: yeah

nmatsakis: agreed on all points

ubsan: the issue was like, if the object lies at the end of the address space

iirc

someting about how this would be UB in C

steveklabnik: ermmm... no

lol

okay

since it was you that brought it up, if you're happy i'm happy

this is something v similar

though the thing i was thinking of was an irc convo

steveklabnik: I didn't bring it up

I've read the C standard

there's nothing about that in tehre

ok

i am very willing to mis-trust my memory

if you have &T, and it's a valid pointer-to-object

err, x: &T

x + 1 is always valid

making it into an array of size one doesn't change anything

T* x

&x[1] is valid

T (*x)[1]

&x[1] is valid

hrmmm

what was I doing

> For the purposes of these operators, a pointer to an object that is not an element of an array behaves the same as a pointer to the first element of an array of length one with the type of the object as its element type.

I would really like if I could have an ASCII C standard

hmm