C++ difference in behavior between if statement setting condition to true and or-ing a condition in loop

Question

I was working on Datalog interpreter for my CS class and I ran into a strange problem where my Rule evaluations took too many passes to complete. After looking at my code I made two modifications in the following that fixed my evaluations to execute in the correct number of passes:

//original form
bool addedFacts = false;
for (X x: xs) {
    addedFacts = addedFacts || set.insert(x).second;
}
//modified form
bool addedFacts = false;
for (X x: xs) {
    if (set.insert(x).second) {
        addedFacts = true;
    }
}

To me these two code structures appear logically equivalent. Is there a reason why one executes correctly and one execute incorrectly/inefficiently? Here is a buildable example of the problem that was occurring:

#include <iostream>
#include <set>
#include <vector>

using std::set;
using std::vector;
using std::cout;
using std::endl;

const int CAP = 100;

class Rule {
public:
    int factor;
    Rule(int factor) {
        this->factor = factor;
    }
    bool evaluateInefficient(set<int>& facts) {
        vector<int> data;
        bool addedFacts = false;
        for (int fact : facts) {
            data.push_back(fact);
        }
        for (int datum : data) {
            int newFact = datum * factor;
            if (newFact < CAP) {
                addedFacts = addedFacts || facts.insert(newFact).second;
            }
        }
        return addedFacts;
    }
    bool evaluate(set<int>& facts) {
        vector<int> data;
        bool addedFacts = false;
        for (int fact : facts) {
            data.push_back(fact);
        }
        for (int datum : data) {
            int newFact = datum * factor;
            if (newFact < CAP) {
                if (facts.insert(newFact).second) {
                    addedFacts = true;
                }
            }
        }
        return addedFacts;
    }
};

int doublyInefficient(vector<Rule>& rules) {
    set<int> facts;
    facts.insert(1);
    bool addedFacts = true;
    int passes = 0;
    while (addedFacts) {
        passes++;
        addedFacts = false;
        for (Rule rule : rules) {
            addedFacts = addedFacts || rule.evaluateInefficient(facts);
        }
    }
    return passes;
}

int singlyInefficient(vector<Rule>& rules) {
    set<int> facts;
    facts.insert(1);
    bool addedFacts = true;
    int passes = 0;
    while (addedFacts) {
        passes++;
        addedFacts = false;
        for (Rule rule : rules) {
            addedFacts = addedFacts || rule.evaluate(facts);
        }
    }
    return passes;
}

int efficient(vector<Rule>& rules) {
    set<int> facts;
    facts.insert(1);
    bool addedFacts = true;
    int passes = 0;
    while (addedFacts) {
        passes++;
        addedFacts = false;
        for (Rule rule : rules) {
            if (rule.evaluate(facts)) {
                addedFacts = true;
            }
        }
    }
    return passes;
}

int main(int argc, char* argv[]) {
    //build the rules
    vector<Rule> rules;
    rules.push_back(Rule(2));
    rules.push_back(Rule(3));
    rules.push_back(Rule(5));
    rules.push_back(Rule(7));
    rules.push_back(Rule(11));
    rules.push_back(Rule(13));
    //Show three different codes that should (in my mind) take the same amount of passes over the rules but don't
    cout << "Facts populated after " << doublyInefficient(rules) << " passes through the Rules." << endl;
    cout << "Facts populated after " << singlyInefficient(rules) << " passes through the Rules." << endl;
    cout << "Facts populated after " << efficient(rules) << " passes through the Rules." << endl;
    getchar();
}

I get the following output when running in debug and release mode (32-bit) on visual studio 2017. To my knowledge the code is not optimized.

Facts populated after 61 passes through the Rules.
Facts populated after 17 passes through the Rules.
Facts populated after 7 passes through the Rules.

Answer 1

A difference arises because of short-circuit evaluation: Consider an expression of the form (expr1 || expr2). Short-circuit means that if expr1 evaluates to true, expression expr2 will not get evaluated at all (cf. this online c++ standard draft, ephasis are mine):

5.15 Logical OR operator

The || operator groups left-to-right. The operands are both contextually converted to bool (Clause [conv]). It returns true if either of its operands is true, and false otherwise. Unlike |, || guarantees left-to-right evaluation; moreover, the second operand is not evaluated if the first operand evaluates to true.

Hence, in your expression addedFacts || set.insert(x).second, from the point where addedFacts becomes true the first time, expression set.insert(x).second will not get executed any more. I suppose that this is the "wrong" behaviour, as your set will not contain the respective xes then.