Syntax and Semantics: Two Different Kinds of Wrong

"The bicycle drank the mountain." Every grammar rule is obeyed. The sentence still means nothing.

Programs go wrong in exactly these two ways. Either the compiler cannot read what you wrote, or it reads it perfectly and does something you did not want. The first kind stops you in one second. The second kind can hide for a month.

Maria asks Bob why his rectangle program prints 8 for a 5 by 3 room. Bob says the compiler gave no errors. That is true, and it is not a defence.

Two kinds of wrong, one sentence each

Syntax is the grammar. Is this order of symbols legal C? The compiler decides, and it is never in doubt.

Semantics is the meaning. Suppose it is legal C. What does it actually do? Only you can answer that, because only you know what you wanted.

So a syntax error is a sentence the machine cannot parse, and a semantic error is a sentence it parses into the wrong instruction.

Syntax errors: eleven messages you will actually see

Every message in this table is real output from GCC 12, the compiler behind the Playground. Read it as a lookup, not as a list to memorise. Find the shape of your message and the row tells you what happened.

A version note. Rows three and four are warnings on GCC 12, which is what the Playground runs. GCC 14 makes both of them errors. If your own compiler refuses a program the Playground builds, this is usually why.

So a syntax error is never a mystery. It is a sentence in a table you will know by heart within a month.

Semantic errors: the compiler has no opinion

Here is Bob's program, in full. It compiles with zero warnings.

#include <stdio.h>

int main(void)
{
    int length = 5;
    int width = 3;
    int area = length + width;   /* legal C, wrong meaning */

    printf("Area = %d\n", area);
    return 0;
}

Area = 8

The area of a 5 by 3 room is 15. The compiler does not know what the word "area" means. It knows you asked it to add two integers, and adding two integers is perfectly legal.

This is the sentence to carry out of this lesson. The compiler guards your grammar. Only you guard your meaning.

So a clean build tells you the machine understood your instructions, and nothing at all about whether they were the right ones.

The = versus == trap

In C, = stores a value and == compares two values. They are one keystroke apart and they mean opposite things.

The reason this is dangerous is that = is not a statement. It is an expression, and an expression has a value. The value of marks = 60 is 60, so you can print it, which is what Example 3 does.

Later, in Module 5, you will put these inside if. There, any value that is not zero counts as true. So if (marks = 90) stores 90, sees a non-zero value, and always runs the branch. The book industry has lost a great deal of money to that single keystroke.

One defence is to write the constant first: if (90 == marks). Slip and type one = there and the compiler refuses, because you cannot store into the number 90. The habit is called a Yoda condition, because it sounds backwards.

So == asks a question, = gives an order, and the compiler is happy with either.

The one-second fix and the one-month fix

A refused build feels like the compiler being difficult. It is the opposite. Every error it prints is a bug you will not be hunting next week.

The gap between syntax and semantics is where real bugs live, and two famous ones show the size of the gap.

In 2014, Apple shipped a TLS bug known as goto fail. One goto fail; line was typed twice. The syntax was perfect, the second copy skipped a certificate check, and connections that should have been rejected were accepted.

In 2008, a Debian maintainer removed two lines from OpenSSL to silence a memory-checking warning. The code still compiled. It also crippled the random number generator, so every key generated on Debian for two years came from a tiny set.

So the compiler catches the cheap mistakes, and the expensive ones are all semantic.

How to read a message: three habits

Read the first error only. One mistake often produces twenty messages, because after the first confusion the compiler is reading your program wrongly. Fix the first, build again.

Check the line above the one named. A missing semicolon is reported where the next statement starts, not where the semicolon should be.

Copy the exact words. Search for expected declaration or statement at end of input, not "weird brace error". Exact words are a debugging tool, which is why the next lesson teaches the name of every symbol.

hello.c : 5 : 12 : error : expected ';' before 'return'
   |      |    |     |            |
   file   line column severity    what the compiler expected

#include <stdio.h>

int main(void)
{
    int length = 5;
    int width = 3;

    printf("Area = %d\n", length + width);
    return 0;
}

Area = 8

#include <stdio.h>

int main(void)
{
    int length = 5;
    int width = 3;

    /* Area of a rectangle is length times width, in square metres. */
    printf("Area = %d\n", length * width);
    return 0;
}

Area = 15

#include <stdio.h>

int main(void)
{
    int marks = 90;

    printf("marks is %d\n", marks);
    printf("marks == 90 gives %d\n", marks == 90);
    printf("marks = 60 gives %d\n", marks = 60);
    printf("marks is now %d\n", marks);
    return 0;
}

marks is 90
marks == 90 gives 1
marks = 60 gives 60
marks is now 60

#include <stdio.h>

int main(void)
{
    int physics = 75;
    int chemistry = 80;
    int total = physics + chemistry;
    int average = total / 2;

    printf("Total: %d\n", total);
    printf("Average: %d\n", average);
    return 0;
}

Total: 155
Average: 77

int marks = 90
printf("%d\n", marks);

int average = total / count;   /* count might be 0 */

if (marks = 90) {
    printf("Top of the class\n");
}

/* outer /* inner */ outer */

#include <stdio.h>

int main(void)
{
    int total = 0;
    int count = 0;

    printf("Average = %d\n", total / count);
    return 0;
}

#include <stdio.h>

int main(void)
{
    int n
    scanf("%d", &n);
    Printf("Double of %d is %d\n", n, n * 2);
    return 0;

#include <stdio.h>

int main(void)
{
    int s;
    scanf("%d", &s);

    printf("Area is %d\n", 4 * s);
    return 0;
}

#include <stdio.h>

int main(void)
{
    int a, b;
    scanf("%d %d", &a, &b);

    printf("%d\n", a = b);
    return 0;
}

Next you will learn the name of every symbol in C, so that the messages in this table become sentences you can read out loud.