Auto merge of #23002 - pnkfelix:fsk-box-place-runway, r=nikomatsakis

Runway for RFC 809 (overloaded box/placement-in) by adding type annotations or explicit calls to `Box::new` where I found it necessary on PR #22086.

I have broken this up into more than one PR because the entire commit chain (see PR #22086) is long, widespread and unwieldy to rebase frequently.

To my knowledge this is not a breaking change.  Also, there is in principle nothing stopping someone from reverting some/all of these annotations, since without the rest of the commit chain in #22086, the associated code would continue to compile.

All I can do is ask: Try to discourage others from removing seemingly "unnecessary" uses of the `Box` type or the `Box::new()` function, until the rest of RFC 809 lands.

1 files changed, 3 insertions, 2 deletions

diff --git a/src/libstd/old_io/timer.rs b/src/libstd/old_io/timer.rs
index 1f2ef50fcae..de5f2141095 100644
--- a/src/libstd/old_io/timer.rs
+++ b/src/libstd/old_io/timer.rs
@@ -15,6 +15,7 @@
 
 // FIXME: These functions take Durations but only pass ms to the backend impls.
 
+use boxed::Box;
 use sync::mpsc::{Receiver, Sender, channel};
 use time::Duration;
 use old_io::IoResult;
@@ -143,7 +144,7 @@ impl Timer {
         let (tx, rx) = channel();
         // Short-circuit the timer backend for 0 duration
         if in_ms_u64(duration) != 0 {
-            self.inner.oneshot(in_ms_u64(duration), box TimerCallback { tx: tx });
+            self.inner.oneshot(in_ms_u64(duration), Box::new(TimerCallback { tx: tx }));
         } else {
             tx.send(()).unwrap();
         }
@@ -204,7 +205,7 @@ impl Timer {
         // not clear what use a 0ms period is anyway...
         let ms = if ms == 0 { 1 } else { ms };
         let (tx, rx) = channel();
-        self.inner.period(ms, box TimerCallback { tx: tx });
+        self.inner.period(ms, Box::new(TimerCallback { tx: tx }));
         return rx
     }
 }