Github user mgaido91 commented on the issue:
https://github.com/apache/spark/pull/19916
Thank you again very much for your comment @kiszk! Results are different
now and they show a difference in performance.
The code I am running is:
```
public class A {
private static Long g;
public static void main(String[] args){
A a = new A();
long start = System.currentTimeMillis();
a.testWithCast();
long end = System.currentTimeMillis();
System.out.println("Time for casting: " + (end-start));
long start1 = System.currentTimeMillis();
a.testWithoutCast();
long end1 = System.currentTimeMillis();
System.out.println("Time without casting: " + (end1-start1));
}
public void testWithCast() {
Object a = new Long(1);
Long b = 0L;
for(long i = 0; i < 1000000000; ++ i){
b += ((Long)a);
}
g = b;
}
public void testWithoutCast() {
Object a = new Long(1);
Long al = (Long) a;
Long b = 0L;
for(long i = 0; i < 1000000000; ++ i){
b += al;
}
g = b;
}
}
```
and the results now are:
```
â janino_tests java A
Time for casting: 3650
Time without casting: 3360
â janino_tests java A
Time for casting: 3682
Time without casting: 3399
â janino_tests java A
Time for casting: 3671
Time without casting: 3347
â janino_tests java A
Time for casting: 3743
Time without casting: 3411
```
Thus the time needed for casting is visible now. I am not sure, then, that
replacing the old method is the right thing to do. WDYT?
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
