break
interacts solely with the closest enclosing loop or switch, whether it be a for
, while
or do .. while
type. It is frequently referred to as a goto in disguise, as all loops in C can in fact be transformed into a set of conditional gotos:
for (A; B; C) D;
// translates to
A;
goto test;
loop: D;
iter: C;
test: if (B) goto loop;
end:
while (B) D; // Simply doesn't have A or C
do { D; } while (B); // Omits initial goto test
continue; // goto iter;
break; // goto end;
The difference is, continue
and break
interact with virtual labels automatically placed by the compiler. This is similar to what return
does as you know it will always jump ahead in the program flow. Switches are slightly more complicated, generating arrays of labels and computed gotos, but the way break works with them is similar.
The programming error the notice refers to is misunderstanding break
as interacting with an enclosing block rather than an enclosing loop. Consider:
for (A; B; C) {
D;
if (E) {
F;
if (G) break; // Incorrectly assumed to break if(E), breaks for()
H;
}
I;
}
J;
Someone thought, given such a piece of code, that G
would cause a jump to I
, but it jumps to J
. The intended function would use if (!G) H;
instead.
与恶龙缠斗过久,自身亦成为恶龙;凝视深渊过久,深渊将回以凝视…