Run A and then B or run C

Question

I'm looking for a control structure to deal with potentially multiple blocks of code or a NONE block. The most basic case is when there are only two conditions, A and B:

+-------+-------+------------+
|   A   |   B   | Blocks Run |
+-------+-------+------------+
| false | false | C          |
| false | true  | B          |
| true  | false | A          |
| true  | true  | A + B      |
+-------+-------+------------+

The best thing I have so far is:

if( A )
{
    // Run Block A
}

if( B )
{
    //Run Block B
}
else if( !A )
{
    //Run Block C
}

No problem but, I want a better control structure if I have a lot of conditions, cause I end up repeating all but the last if check. — Jonathan Mee, Feb 07 '14 at 12:27
I had considered that but I couldn't come up with a way to make it work well. — Jonathan Mee, Feb 07 '14 at 12:29
For clarity, why do you have to avoid checking conditions more than once? — nhgrif, Feb 07 '14 at 12:35

alk · Answer 1 · 2014-02-07T12:48:13.290

4

If it's all about minimising comparsions using a table is the only way to go.

switch ((B << 1) | A)
{
  case 0:
    // C
    break;

  case 0b01: // case 1:
    // A
    break;

  case 0b10: // case 2:
    // B
    break;

  case 0b11: // case 3:
    // A
    // B
    break;

  default:
    // Shouldn't arrive here.
    break;
}

edited Feb 07 '14 at 12:48

answered Feb 07 '14 at 12:38

alk

69,737
10
105
255

2

Nice solution, an additional suggestion: make `case 0b11` fall through to `case 0b01` so you don't have to write block `B` twice. – DarkDust Feb 07 '14 at 12:41
I had considered this as well... It suffers from code replication similarly to @Dark Dust's answer. And would become unwieldy for a higher number of conditionals. – Jonathan Mee Feb 07 '14 at 12:41
@JonathanMee: Shifting the blocks (A, B, C) to functions, I wouldn't consider this messy. – alk Feb 07 '14 at 12:44
@DarkDust: The fall through doesn't make this more readable. I'd leave the optimisation to the compiler. – alk Feb 07 '14 at 12:45

Andrejovich · Accepted Answer · 2014-02-07T12:33:45.560

3

For clarity, I'm going to rename your conditions P and Q (as is the convention in propositional logic)

if( P || Q)
{
    P && A();
    Q && B();
}
else
{
    C();
}

Edit:

adding more checks is rather easy:

if( P || Q || R || S)
{
    P && A();
    Q && B();
    R && D();
    S && E();
}
else
{
    C();
}

Edit, after comment:

Well how about this.

int runC = 1;
P && (runC=0, A());
Q && (runC=0, B());
R && (runC=0, D());
S && (runC=0, E());
runC && C();

edited Feb 07 '14 at 12:33

answered Feb 07 '14 at 12:27

Andrejovich

544
2
13

This still does what I want to avoid: duplicates the if statements, I test each one _twice_. – Jonathan Mee Feb 07 '14 at 12:30
Ah I see you suggest using a flag which would be set on entry to any block. – Jonathan Mee Feb 07 '14 at 12:38
1

The third answer here is the useful one the first two answers, have a worst case evaluation of all variables twice. – Jonathan Mee Feb 07 '14 at 16:43
+1 for the last one... `P`, `Q`.. etc can be functions with side-effect. :-) – anishsane Mar 10 '14 at 12:19

score 1 · Answer 3 · answered Feb 07 '14 at 12:33

If you just want to evaluate each condition once:

if (a) {
  A();
  if (b) {
    B();
  }
} else if (b) {
  B();
} else {
  C()
}

I don't think it's much better than your code, though. While it does evaluate the condition only once, the code blocks to execute are now duplicated. As far as I can see, you'll either have to evaluate a condition twice or a write a code block twice. In that case, I'd prefer evaluating the conditions and saving them to variables, then go with your example: duplicating blocks of code is worse than reading a variable twice (most of the time, it's more important to have code that you can understand and maintain instead of trying to "optimize" it when you haven't measured that it's a bottleneck).

For two conditions this is probably the best solution. So I'll probably end up accepting it, but I don't think it would be a good option for say five conditions. — Jonathan Mee, Feb 07 '14 at 12:36

score 0 · Answer 4 · answered Feb 07 '14 at 12:40

0

You want to run block A when condition A, followed by block B when condition B and so on. With one exception: When no criteria is met then execute the NONE block. Right? Then:

if( !A && !B && !C && ...)
{
  run NONE-BLOCK;
}
else
{
  if( A ) run A-BLOCK;
  if( B ) run B-BLOCK;
  if( C ) run C-BLOCK;
  ...
}

answered Feb 07 '14 at 12:40

Thorsten Kettner

89,309
7
49
73

Worst case this would test all variables twice, best case it's just a little worse than the control structure in the question. – Jonathan Mee Feb 07 '14 at 12:43
@ThorstenKettner: So you probaby do not answer the question, don't you? – alk Feb 07 '14 at 12:58
I think I do. Jonathan Mee is "looking for a control structure to deal with potentially multiple blocks of code or a NONE block". His approach is efficient, but lacks readability (especially when it is not only two conditions and blocks to deal with). So the answer how to improve it, is how to enhance readability, as this is the only flaw in his code. – Thorsten Kettner Feb 07 '14 at 13:06

Run A and then B or run C

4 Answers4

Linked