From a91ad60158647c1f6a89b9c01915279ce9314a65 Mon Sep 17 00:00:00 2001 From: Peter Todd Date: Sat, 18 May 2019 22:27:33 -0400 Subject: [PATCH] =?UTF-8?q?Make=20clear=20that=20status=20quo=20=E2=89=A0?= =?UTF-8?q?=20guarantee?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- src/test/run-pass/union/union-nonzero.rs | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/src/test/run-pass/union/union-nonzero.rs b/src/test/run-pass/union/union-nonzero.rs index 2108cc3a4ea..bd84b46bf3d 100644 --- a/src/test/run-pass/union/union-nonzero.rs +++ b/src/test/run-pass/union/union-nonzero.rs @@ -6,8 +6,9 @@ // For example, if a union `U` can contain both a `&T` and a `*const T`, there's definitely no // bit-value that an `Option` could reuse as `None`; this test makes sure that isn't done. // -// Secondly, this tests the status quo to not apply such optimizations to types containing unions -// even if they're theoretically possible. (discussion: https://github.com/rust-lang/rust/issues/36394) +// Secondly, this tests the status quo (not a guarantee; subject to change!) to not apply such +// optimizations to types containing unions even if they're theoretically possible. (discussion: +// https://github.com/rust-lang/rust/issues/36394) // // Notably this nails down part of the behavior that `MaybeUninit` assumes: that a // `Option>` does not take advantage of non-zero optimization, and thus is a safe