Refine PR #642 (#665)

src/dimension/dynindeximpl.rs

@@ -49,7 +49,7 @@ impl<T: Copy + Zero> IxDynRepr<T> {
     pub fn copy_from(x: &[T]) -> Self {
         if x.len() <= CAP {
             let mut arr = [T::zero(); CAP];
-            arr[..x.len()].clone_from_slice(&x[..]);
+            arr[..x.len()].copy_from_slice(&x[..]);
             IxDynRepr::Inline(x.len() as _, arr)
         } else {
             Self::from(x)

src/lib.rs

@@ -1538,14 +1538,11 @@ where
         F: FnMut(&mut A),
     {
         if let Some(slc) = self.as_slice_memory_order_mut() {
-            // FIXME: Use for loop when slice iterator is perf is restored
-            for x in slc.iter_mut() {
-                f(x);
+            slc.iter_mut().for_each(f);
+        } else {
+            for row in self.inner_rows_mut() {
+                row.into_iter_().fold((), |(), elt| f(elt));
             }
-            return;
-        }
-        for row in self.inner_rows_mut() {
-            row.into_iter_().fold((), |(), elt| f(elt));
         }
     }
 

src/numeric_util.rs

@@ -6,8 +6,6 @@
 // option. This file may not be copied, modified, or distributed
 // except according to those terms.
 
-// https://github.com/rust-ndarray/ndarray/pull/642#discussion_r296074711
-#![allow(clippy::needless_range_loop)]
 use std::cmp;
 
 use crate::LinalgScalar;
@@ -51,12 +49,11 @@ where
 
     // make it clear to the optimizer that this loop is short
     // and can not be autovectorized.
-    // https://github.com/rust-ndarray/ndarray/pull/642#discussion_r296337112
-    for i in 0..xs.len() {
+    for (i, x) in xs.iter().enumerate() {
         if i >= 7 {
             break;
         }
-        acc = f(acc.clone(), xs[i].clone())
+        acc = f(acc.clone(), x.clone())
     }
     acc
 }
@@ -103,14 +100,11 @@ where
     sum = sum + (p2 + p6);
     sum = sum + (p3 + p7);
 
-    for i in 0..xs.len() {
+    for (i, (&x, &y)) in xs.iter().zip(ys).enumerate() {
         if i >= 7 {
             break;
         }
-        unsafe {
-            // get_unchecked is needed to avoid the bounds check
-            sum = sum + xs[i] * *ys.get_unchecked(i);
-        }
+        sum = sum + x * y;
     }
     sum
 }