I was wondering what the performance difference would be between using arrayByAddingObject and creating a mutable copy and then adding one element.
I found that it doesn't matter.
Here's what I used
- (IBAction)doTest
{
for(int n = 0; n < 7; n++){
[self runTestWithExponent:n];
}
NSLog(@"done");
}
- (void)runTestWithExponent:(double)exponent
{
int arraySize = pow(10.0, exponent);
NSMutableArray *originalMutableArray = [NSMutableArray arrayWithCapacity:arraySize];
for(int i = 0; i < arraySize; i++){
[originalMutableArray addObject:@(i)];
}
NSArray *originalArray = [NSArray arrayWithArray:originalMutableArray];
originalMutableArray = nil;
//test first time
NSDate *now = [NSDate date];
NSArray *newArray = [originalArray arrayByAddingObject:@(-1)];
NSTimeInterval time1 = [[NSDate date] timeIntervalSinceDate:now];
newArray = nil;
//test second time
now = [NSDate date];
NSMutableArray *mutable = [originalArray mutableCopy];
[mutable addObject:@(-1)];
NSTimeInterval time2 = [[NSDate date] timeIntervalSinceDate:now];
mutable = nil;
NSString *winner = (time1 == time2) ? @"same" : ((time1 < time2) ? @"arrayByAdding" : @"mutable");
NSLog(@"%i : %f --- %f : %@ %f%%", arraySize, time1, time2, winner, (time1/time2 * 100));
}
As you can see, I test different array sizes. Sizes of 1, 10, 100 ... 1,000,000. I found regardless of size, the times are very similar to each other. For any given length, sometimes one method is faster, and others, the other. I'm guessing internally they do the same thing.
Here's the output of three runs:
1 : 0.000026 --- 0.000034 : arrayByAdding 76.491228%
10 : 0.000011 --- 0.000011 : same 100.000000%
100 : 0.000021 --- 0.000024 : arrayByAdding 87.344913%
1000 : 0.000228 --- 0.000260 : arrayByAdding 87.689133%
10000 : 0.001458 --- 0.001406 : mutable 103.696638%
100000 : 0.015396 --- 0.015625 : arrayByAdding 98.534393%
1000000 : 0.158018 --- 0.162849 : arrayByAdding 97.033438%
done
1 : 0.000011 --- 0.000012 : arrayByAdding 92.039801%
10 : 0.000012 --- 0.000014 : arrayByAdding 85.531915%
100 : 0.000020 --- 0.000025 : arrayByAdding 79.952267%
1000 : 0.000185 --- 0.000144 : mutable 128.435430%
10000 : 0.001397 --- 0.001437 : arrayByAdding 97.216807%
100000 : 0.014448 --- 0.014132 : mutable 102.235803%
1000000 : 0.145622 --- 0.149862 : arrayByAdding 97.170746%
done
1 : 0.000013 --- 0.000012 : mutable 107.920792%
10 : 0.000011 --- 0.000013 : arrayByAdding 84.862385%
100 : 0.000025 --- 0.000029 : arrayByAdding 86.036961%
1000 : 0.000165 --- 0.000135 : mutable 122.207506%
10000 : 0.001547 --- 0.001470 : mutable 105.242884%
100000 : 0.014817 --- 0.014337 : mutable 103.347954%
1000000 : 0.146554 --- 0.148468 : arrayByAdding 98.710857%
done
Keep in mind that this was only tested for adding one element. I suspect that (for example) adding 500 elements to the 1,000 sized array will probably be faster using the second method, but I leave such experiments to someone else.