change semantics of `@memcpy` and `@memset`

Now they use slices or array pointers with any element type instead of
requiring byte pointers.

This is a breaking enhancement to the language.

The safety check for overlapping pointers will be implemented in a
future commit.

closes #14040

1 files changed, 1 insertions, 2 deletions

diff --git a/lib/std/multi_array_list.zig b/lib/std/multi_array_list.zig
index 838965f8b1..063e2e5620 100644
--- a/lib/std/multi_array_list.zig
+++ b/lib/std/multi_array_list.zig
@@ -360,11 +360,10 @@ pub fn MultiArrayList(comptime T: type) type {
                     if (@sizeOf(field_info.type) != 0) {
                         const field = @intToEnum(Field, i);
                         const dest_slice = self_slice.items(field)[new_len..];
-                        const byte_count = dest_slice.len * @sizeOf(field_info.type);
                         // We use memset here for more efficient codegen in safety-checked,
                         // valgrind-enabled builds. Otherwise the valgrind client request
                         // will be repeated for every element.
-                        @memset(@ptrCast([*]u8, dest_slice.ptr), undefined, byte_count);
+                        @memset(dest_slice, undefined);
                     }
                 }
                 self.len = new_len;