MISRA rule 3.1 comments may be too restrictive

[Rob Gordon]

HI! First of all, thank you for implementing MISRA C checks in cppcheck. We just started applying these checks to our code as a request from a customer. My past experience has shown that most of these checks will provide real benefit and I have seen/fixed many of them in my career.

It appears that the way rule 3.1 is applied by cppcheck is too restrictive (cppcheck 1.84). Granted, I believe the rule needs some clarification by the committee. The example code given:

x=y // /* non-compliant
 +x
   // */  
 ;  
 //// should be compliant?

The first sequence of // / is reported as non-compliant by cppcheck, good.
The second sequence of // / is not reported but is not explicitly addressed by MISRA, not an issue right now.
But the third sequence (added by me for testing) is multiple // sequences together, ie ////. cppcheck complains that these are non-compliant. But there is a stated exception for this in rule 3.1. As a comparison, we have access to the Altium TASKING compiler and it allows the multiple // sequence.

I believe cppcheck should address the exception stated in the MISRA doc. An updated check for the misra.py script would look like:

def misra_3_1(rawTokens):
    for token in rawTokens:
        if token.str.startswith('/*'):
            if '//' in token.str[2:] or '/*' in token.str[2:]:
                reportError(token, 3, 1)
        elif token.str.startswith('//'):
            if '/*' in token.str[2:]:
                reportError(token, 3, 1)

Cheers,
Rob

[Daniel Marjamäki]

Thanks! I agree. I made a fix for this.