if on the same line

Reasons I prefer

// Case1
if (A) {
    B;
}

over

// Case2
if (A) B;

1. Case1 is more debugger friendly - it is trivial to set a breakpoint on B. In Case2 if a breakpoint is set on that line, then the debugger will stop before A is evaluated. Yes, it is possible to set a condition on the breakpoint to stop only if A is true, but that is more tedious and it means A will be evaluated two times which is problematic if A has side effects.

2. Case1 is more diff friendly in case it has to be expanded:

// Case1
 if (A) {
+    N;
     B;
 }

vs

// Case2
-if (A) B;
+if (A) {
+    N;
+    B;
+}

especially if A and B are complex expressions - the reviewer has to compare whether each one has been changed.

For example (this is real code):

-        if (m_shift < n || m_shift >= std::numeric_limits<uint64_t>::max() || m_shift < std::numeric_limits<I>::min() || m_shift > std::numeric_limits<I>::max()) throw std::ios_base::failure("differential value overflow");
+        if (m_shift < n || m_shift >= std::numeric_limits<uint64_t>::max() || m_shift < std::numeric_limits<I>::min() || m_shift > std::numeric_limits<I>::max()) {
+            foo;
+            throw std::ios_base::failure("differential value overflow");
+        }

Was the if condition changed?

What about this:

         if (m_shift < n || m_shift >= std::numeric_limits<uint64_t>::max() || m_shift < std::numeric_limits<I>::min() || m_shift > std::numeric_limits<I>::max()) {
+            foo;
             throw std::ios_base::failure("differential value overflow");
         }