bytes: improve WriteRune performance

Remove the runeBytes buffer and write the utf8 encoding directly
to the internal buf byte slice.

name         old time/op   new time/op   delta
WriteRune-4   80.5µs ± 2%   57.1µs ± 2%  -29.06%  (p=0.000 n=20+20)

name         old speed     new speed     delta
WriteRune-4  153MB/s ± 2%  215MB/s ± 2%  +40.96%  (p=0.000 n=20+20)

Change-Id: Ic15f6e2d6e56a3d15c74f56159e2eae020ba73ba
Reviewed-on: https://go-review.googlesource.com/28816Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

src/bytes/buffer.go

@@ -17,7 +17,6 @@ import (
 type Buffer struct {
 	buf       []byte   // contents are the bytes buf[off : len(buf)]
 	off       int      // read at &buf[off], write at &buf[len(buf)]
-	runeBytes [utf8.UTFMax]byte // avoid allocation of slice on each call to WriteRune
 	bootstrap [64]byte // memory to hold first slice; helps small buffers avoid allocation.
 	lastRead  readOp   // last read operation, so that Unread* can work correctly.
 }
@@ -246,8 +245,10 @@ func (b *Buffer) WriteRune(r rune) (n int, err error) {
 		b.WriteByte(byte(r))
 		return 1, nil
 	}
-	n = utf8.EncodeRune(b.runeBytes[0:], r)
-	b.Write(b.runeBytes[0:n])
+	b.lastRead = opInvalid
+	m := b.grow(utf8.UTFMax)
+	n = utf8.EncodeRune(b.buf[m:m+utf8.UTFMax], r)
+	b.buf = b.buf[:m+n]
 	return n, nil
 }
 

src/bytes/buffer_test.go

@@ -514,6 +514,19 @@ func TestBufferGrowth(t *testing.T) {
 	}
 }
 
+func BenchmarkWriteRune(b *testing.B) {
+	const n = 4 << 10
+	const r = '☺'
+	b.SetBytes(int64(n * utf8.RuneLen(r)))
+	buf := NewBuffer(make([]byte, n*utf8.UTFMax))
+	for i := 0; i < b.N; i++ {
+		buf.Reset()
+		for i := 0; i < n; i++ {
+			buf.WriteRune(r)
+		}
+	}
+}
+
 // From Issue 5154.
 func BenchmarkBufferNotEmptyWriteRead(b *testing.B) {
 	buf := make([]byte, 1024)