Github user original-brownbear commented on a diff in the pull request:
https://github.com/apache/spark/pull/19180#discussion_r137951583
--- Diff:
common/unsafe/src/main/java/org/apache/spark/unsafe/types/UTF8String.java ---
@@ -1097,8 +1101,21 @@ public UTF8String copy() {
@Override
public int compareTo(@Nonnull final UTF8String other) {
int len = Math.min(numBytes, other.numBytes);
- // TODO: compare 8 bytes as unsigned long
- for (int i = 0; i < len; i ++) {
+ int words = len / Longs.BYTES;
+ long roffset = other.getBaseOffset();
+ Object rbase = other.getBaseObject();
+ for (int i = 0; i < words * Longs.BYTES; i += Longs.BYTES) {
+ long left = getLong(base, offset + i);
+ long right = getLong(rbase, roffset + i);
+ if (left != right) {
+ if (!IS_LITTLE_ENDIAN) {
--- End diff --
@kiszk I think it'll result in exactly the same native code after JIT
inlining the `static final boolean` check => I think this is cleaner since we
don't need to duplicate the loop logic.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
