Return to Answer

##Style

##Algorithm

##Update - Performance

Style

Algorithm

Update - Performance

• it won't recurse for negatives, but instead do some more complicated sign-juggling
• it keeps everything as primitives (chars, etc.) instead of Strings
• it pre-computes the output sizes, etc.

Task NumberPad -> OP: (Unit: MILLISECONDS)
 Count : 10000 Average : 0.2202
 Fastest : 0.1960 Slowest : 9.8499
 95Pctile : 0.2927 99Pctile : 0.4620
 TimeBlock : 0.265 0.222 0.222 0.215 0.218 0.217 0.217 0.208 0.206 0.211
 Histogram : 9769 218 7 1 4 1
Task NumberPad -> RL: (Unit: MILLISECONDS)
 Count : 10000 Average : 0.3524
 Fastest : 0.3127 Slowest : 15.0630
 95Pctile : 0.5521 99Pctile : 0.6260
 TimeBlock : 0.416 0.368 0.360 0.340 0.366 0.347 0.330 0.334 0.328 0.335
 Histogram : 9900 88 6 2 3 1
Task NumberPad -> RLP: (Unit: MILLISECONDS)
 Count : 10000 Average : 0.1866
 Fastest : 0.1683 Slowest : 5.9859
 95Pctile : 0.2280 99Pctile : 0.4241
 TimeBlock : 0.218 0.189 0.193 0.181 0.185 0.185 0.177 0.178 0.179 0.182
 Histogram : 9826 165 3 2 2 2

• it won't use recursion for negatives, but instead do some more complicated sign-juggling
• it keeps everything as primitives (chars, etc.) instead of Strings
• it pre-computes the output sizes, etc.

Task NumberPad -> OP: (Unit: MILLISECONDS)
 Count : 10000 Average : 0.2202
 Fastest : 0.1960 Slowest : 9.8499
 95Pctile : 0.2927 99Pctile : 0.4620
 TimeBlock : 0.265 0.222 0.222 0.215 0.218 0.217 0.217 0.208 0.206 0.211
 Histogram : 9769 218 7 1 4 1
Task NumberPad -> RL: (Unit: MILLISECONDS)
 Count : 10000 Average : 0.3524
 Fastest : 0.3127 Slowest : 15.0630
 95Pctile : 0.5521 99Pctile : 0.6260
 TimeBlock : 0.416 0.368 0.360 0.340 0.366 0.347 0.330 0.334 0.328 0.335
 Histogram : 9900 88 6 2 3 1
Task NumberPad -> RLP: (Unit: MILLISECONDS)
 Count : 10000 Average : 0.1866
 Fastest : 0.1683 Slowest : 5.9859
 95Pctile : 0.2280 99Pctile : 0.4241
 TimeBlock : 0.218 0.189 0.193 0.181 0.185 0.185 0.177 0.178 0.179 0.182
 Histogram : 9826 165 3 2 2 2

##Update - Performance

Since the question was edited to include performance as a criteria, then the following code will do essentially the same as above, but with some performance tweaks:

• it won't recurse for negatives, but instead do some more complicated sign-juggling
• it keeps everything as primitives (chars, etc.) instead of Strings
• it pre-computes the output sizes, etc.

It is still referenced against the right-side of the output, and has the same basic concepts, just... faster. About twice as fast, and about 20% faster than the OP's...

public static String neatifyRLP(final long val, final char pad, final int span) {
 final String raw = Long.toString(val);
 final int signlen = val < 0 ? 1 : 0;
 final int digitlen = raw.length() - signlen;
 final char[] out = new char[signlen + digitlen + (digitlen - 1) / span];
 int pos = out.length - 1;
 int src = raw.length() - 1;
 final int ospan = span + 1;
 while (pos >= signlen) {
 out[pos] = (out.length - pos) % ospan == 0 ? pad : raw.charAt(src--);
 pos--;
 }
 if (val < 0) {
 out[0] = '-';
 }
 return new String(out);
}

I am not at all certain that it is more readable, but, it is not horrible either.

Here are the performance numbers based on neatifying 1000 random long values:

Task NumberPad -> OP: (Unit: MILLISECONDS)
 Count : 10000 Average : 0.2202
 Fastest : 0.1960 Slowest : 9.8499
 95Pctile : 0.2927 99Pctile : 0.4620
 TimeBlock : 0.265 0.222 0.222 0.215 0.218 0.217 0.217 0.208 0.206 0.211
 Histogram : 9769 218 7 1 4 1
Task NumberPad -> RL: (Unit: MILLISECONDS)
 Count : 10000 Average : 0.3524
 Fastest : 0.3127 Slowest : 15.0630
 95Pctile : 0.5521 99Pctile : 0.6260
 TimeBlock : 0.416 0.368 0.360 0.340 0.366 0.347 0.330 0.334 0.328 0.335
 Histogram : 9900 88 6 2 3 1
Task NumberPad -> RLP: (Unit: MILLISECONDS)
 Count : 10000 Average : 0.1866
 Fastest : 0.1683 Slowest : 5.9859
 95Pctile : 0.2280 99Pctile : 0.4241
 TimeBlock : 0.218 0.189 0.193 0.181 0.185 0.185 0.177 0.178 0.179 0.182
 Histogram : 9826 165 3 2 2 2