How to Fix MISRA C++ Rule 0-1-4

Question

The following code violates the MISRA C++ rule 0-1-4:

for (auto &a : b) {
    ... // The variable a is used only in the for condition.
}

Rule: A project shall not contain non-volatile POD variables having only one use. Variable 'a' is used only once, that is, during initialization.

What I tried:

for (const auto &a : b) {
    ... // The variable a is used only in the for condition.
}

But that was not the solution.

Does anyone have an idea how I can fix it?

Perhaps request an exemption when a coding standard gets in the way? Tools should be working for you, not the other way around. — n. m. could be an AI, May 11 '22 at 17:12
@n.1.8e9-where's-my-sharem. Do you think I should ignore this violation? — KplnMoon, May 11 '22 at 17:24
If your project rules allow you to choose which violation to ignore, then I'd say feel free to ignore it and suppress the message. Though if `(void)a;` trick works, feel free to use it too. I assumed it would violate some other rule. — n. m. could be an AI, May 11 '22 at 17:51
The solution below does not violate any other rule. Thanks anyway for your tip. Since this MISRA rule is from the "**required**" category, you can't ignore it (as I understand it). — KplnMoon, May 11 '22 at 18:16
You can't use C++11 and MISRA C++ together, it's completely senseless. It's like having a corn flakes factory and placing one piece of nitroglycerine in each packet of corn flakes, then worry about if the nitroglycerine-placing machine is safe to use for the operator. Who cares, since you are, by design, counting on each of your products to explode in your customer's face? — Lundin, May 12 '22 at 08:05

Bathsheba · Accepted Answer · 2022-05-11T17:14:59.060

2

Alas the current C++ grammar requires you to declare a variable when using the range-for form of the for loop:

for (auto& : b) {

is not allowed, despite it having potential applications (such as computing the number of elements in a container).

Writing

a;

or

(void)a;

in the loop body might work depending on the type. This would fool the static analyser and you can hope the expression gets compiled out.

edited May 11 '22 at 17:14

answered May 11 '22 at 16:58

Bathsheba

Thanks for the quick reply! :) That fixed it. But the new line violates another rule: "There shall be no dead code. Expression has no effect" – KplnMoon May 11 '22 at 17:11
@KplnMoon: I've added another trick. – Bathsheba May 11 '22 at 17:15
Wow! **(void)a;** does not cause any additional error and solves the problem. Thanks again! – KplnMoon May 11 '22 at 17:22
1

@KplnMoon that is a verbose **shut-up** to the compiler. – Red.Wave May 11 '22 at 18:08
@Red.Wave So you don't think it should be solved that way? – KplnMoon May 11 '22 at 18:18
@KplnMoon it's normally used as a patch. You have not used the object, yet it has a name. – Red.Wave May 11 '22 at 18:21
`#include ` for c++20: `std::ranges::for_each(b,[&](auto const &/*a*/){/*a has no name*/});` or `std::for_each(begin(b),end(b),[&](auto const&){/*a has no name again*/});` – Red.Wave May 11 '22 at 18:26
If there is a `break` then `find_if` works. `return true;` to `break` and `return false;` to `continue` with next element. These 2 library solutions are also imune to an alledged bug in range-based `for`. – Red.Wave May 11 '22 at 18:40

1 Answers1