Add copy_to_slice #331
Add copy_to_slice #331
Conversation
calebzulawski
requested review from
thomcc,
programmerjake and
workingjubilee
calebzulawski
force-pushed
the
to_slice
branch
from
e449630 to
0fd7c8e
Compare
| // Safety: | ||
| // - We've checked the length is sufficient | ||
| // - `T` and `Simd<T, N>` are Copy types. | ||
| unsafe { slice.as_mut_ptr().cast::<Self>().write_unaligned(self) } |
There was a problem hiding this comment.
what happens if the simd type has padding (e.g. for non-power-of-2 types where #319 isn't fixed yet)?
imho we should add an assert that the simd type and corresponding array type have the same size, otherwise write_unaligned is permitted to overwrite after-the-end elements with undef padding -- unsound. panicking is better than UB and the assert will never fire once #319 is fixed.
There was a problem hiding this comment.
also, you might want an explicit Copy bound since we might add Simd<&mut _, _> support -- which isn't Copy.
There was a problem hiding this comment.
Good point about the padding. I think I'll leave out the Copy bound for because basically every one of Simd's associated functions fail if the element isn't Copy, that would probably need a much more significant overhaul.
Closes #324