Uncommon but useful GCC command line options - part 2

If you recall, in the beginning of the first part of this tutorial series, I briefly mentioned that the -Wall option that developers generally use for generating warnings doesn't cover some specific warnings. If you aren't aware of these warnings and have no idea on how to enable them, worry not, as we'll be explaining all that in detail in this article.

Aside from this, we'll also be covering a gcc warning option related to floating point variables, as well as how to better manage the gcc command line options if the list grows large.

But before we move ahead, please keep in mind that all the examples, command, and instructions mentioned in this tutorial have been tested on Ubuntu 16.04 LTS, and the gcc version that we've used is 5.4.0.

Enable warnings that aren't covered by -Wall

While the -Wall option of the gcc compiler covers most of the warning flags, there are some that remain disabled. To enable them, use the -Wextra option.

For example, take a look at the following code:

#include <stdio.h>
#include <stdlib.h>
int main()
{
    int i=0;
    /* ...
       some code here 
       ...
    */
    
    if(i);
        return 1;
     return 0; 
}

I accidentally put a semicolon after the 'if' condition. Now, when the code was compiled using the following gcc command, no warning was produced.

gcc -Wall test.c -o test

But when the -Wextra option was used:

gcc -Wall -Wextra test.c -o test

A warning was produced:

test.c: In function ‘main’:
test.c:10:8: warning: suggest braces around empty body in an ‘if’ statement [-Wempty-body]
 if(i);

As clear from the warning shown above, the -Wextra option internally enabled the -Wempty-body flag, which in turned detected the suspicious code and issued the warning. Here is the complete list of warning flags enabled by this option:

-Wclobbered, -Wempty-body, -Wignored-qualifiers, -Wmissing-field-initializers, -Wmissing-parameter-type (C only), -Wold-style-declaration (C only), -Woverride-init, -Wsign-compare, -Wtype-limits, -Wuninitialized, -Wunused-parameter (only with -Wunused or -Wall), and -Wunused-but-set-parameter (only with -Wunused or -Wall).

If you want to learn what the above mentioned flags do, head to gcc's man page.

Moving on, the -Wextra option also issues warnings in the following cases:

A pointer is compared against integer zero with <, <=, >, or >=
(C++ only) An enumerator and a non-enumerator both appear in a
conditional expression.
(C++ only) Ambiguous virtual bases.
(C++ only) Subscripting an array that has been declared
register.
(C++ only) Taking the address of a variable that has been
declared register.
(C++ only) A base class is not initialized in a derived class's
copy constructor.

Enable warning for floating point values in equity comparisons

As you might already know, one should never test for the exact equality of floating-point values (didn't know this - read the floating-point comparison-related FAQ here). But even if you accidentally do this, does the gcc compiler throw any warning or error? Let's check out:

Here's a code that compares floating point variables using == operator:

#include<stdio.h>

void compare(float x, float y)
{
    if(x == y)
    {
        printf("\n EQUAL \n");
    }
}


int main(void)
{
    compare(1.234, 1.56789);
 
    return 0; 
}

And here's the gcc command (containing both -Wall and -Wextra options) used to compile this code:

gcc -Wall -Wextra test.c -o test

Sadly, the above command doesn't produce any warning related to floating point comparison. A quick look at GCC's man page reveals that there exists a dedicated option -Wfloat-equal that should be used in these scenarios.

Here's the command containing this option:

gcc -Wall -Wextra -Wfloat-equal test.c -o test

And following is the output it produced:

test.c: In function ‘compare’:
test.c:5:10: warning: comparing floating point with == or != is unsafe [-Wfloat-equal]
 if(x == y)

So as you can see in the output above, the -Wfloat-equal option forced gcc to produce warning related to floating comparison.

Here is what the gcc man page says about this option:

The idea behind this is that sometimes it is convenient (for the programmer) to consider floating-point values as approximations to infinitely precise real numbers. If you are doing this, then you
need to compute (by analyzing the code, or in some other way) the maximum or likely maximum error that the computation introduces, and allow for it when performing comparisons (and when producing
output, but that's a different problem). In particular, instead of testing for equality, you shouldcheck to see whether the two values have ranges that overlap; and this is done with the relational operators, so equality comparisons are probably mistaken.

How to better manage gcc command line options

If the list of command line options that you are using in your gcc command is becoming larger and difficult to manage, then you can put it in a text file, and pass that file's name as an argument to the gcc command. For this, you'll have to use the @file command line option.

For example, if following is your gcc command:

gcc -Wall -Wextra -Wfloat-equal test.c -o test

Then you can put the three warnings-related options in a file named, say 'gcc-options' :

$ cat gcc-options 
-Wall -Wextra -Wfloat-equal

And your gcc command becomes less cluttered and easy to manage:

gcc @gcc-options test.c -o test

Here's what the gcc man page says about @file:

Read command-line options from file. The options read are inserted in place of the original @file option. If file does not exist, or cannot be read, then the option will be treated literally, and not removed.

Options in file are separated by whitespace. A whitespace character may be included in an option by surrounding the entire option in either single or double quotes. Any character (including a backslash) may be included by prefixing the character to be included with a backslash. The file may itself contain additional @file options; any such options will be processed recursively.

Conclusion

So we covered a total of 5 uncommon but useful gcc command line options in this tutorial series: -save-temps, -g, -Wextra, -Wfloat-equal, and @file. Do spend time practicing each of them and don't forget to go through all the details that the gcc man page offers about them.

Do you know or use other such useful gcc command line options, and want to share them with the world? Leave all the details in comments below.

Log in or Sign up