Is 'X = someFunction() + 2` a statement or an expression?

Question

I read that an expression is anything that gives some value, like 2 + X, while a statement is any instruction to the computer to execute something, like print("hi").

What about the following line of code?

X = someFunction() + 2

someFunction() returns some numerical value (I think a lot of languages wouldn't compile this code if it didn't), and thus someFunction() + 2 is 'something that yields some value' - aka an expression.

But, someFunction() is code to be executed, thus a statement.

My question:

There are often lines of code that equal some value, but are also an instruction to be executed. What are these lines of code considered?

Answer 1

In certain computer languages called "functional languages", everything--including the code that prints "hi"--is an expression. At the other extreme, you can write code in machine language (so you, not a compiler, are deciding exactly what sequence of bytes should compose the executable program), and at that level practically everything (even adding 2 to something) is an "instruction to the computer to execute something".

I've used a lot of different computer languages, and a far as I can recall, in each case there was documentation somewhere defining what makes a statement in that particular language (if indeed the language even has a concept of "statement"). The definition is based on syntax, not so much on what the code does.

For example, in C or C++, if you write

{ x + 2; }

then technically the "x + 2;" is a statement. It is a useless statement that doesn't do anything, but syntactically, it is a statement nevertheless. In fact, one way to write a statement in C is to just append a semicolon to an expression (http://msdn.microsoft.com/en-us/library/1t054cy7.aspx). You don't even need the expression; a semicolon by itself can be a statement (http://msdn.microsoft.com/en-us/library/h7zyw61x.aspx).

By the way, in C++, the '+' in an expression such as (x + 2) may actually be a function call. So if you say anything that calls a function is a statement, then (x + 2) would be, or at least could be, a statement in C++. But I don't know any authority who defines it that way.

Answer 2

It varies by language, but ultimately: a statement is anything you can't embed inside another (simple, i.e. not a block) statement.

In C, your example is an expression, because you can do this:

while (X = someFunction() + 2) {
    // ...
}

But in Python, the same thing is a syntax error, because = can only be a statement:

# nope!
while X = someFunction() + 2:
    pass

In most languages, any expression can also be used as a statement by itself, though this may or may not be useful.

Calling a statement an "instruction to execute something" is a poor way to think about it, though. All code is an instruction to execute something.

A statement is more like a single complete thought. It's really just part of the syntax; depending on the language/compiler/runtime, a statement or expression may become very many machine instructions, or several statements might be reduced to just one instruction.

Answer 3

tldr; It is a statement when it is parsed as statement, and an expression when it is parsed as an expression. The rules of which depend upon the particular language in question.

Expressions and statements should not be confused with "what actually happens" underneath, but merely as describing the syntax constructs of a language's grammar.

Because the grammar and parsing rules [generally] depend on the program as a whole, taking part of an expression and using it as a statement, where such is allowed, does not indicate that it is a statement, much less when it appears in an expression context.

---

As for the particular example given, it depends on programming language and where the construct appears. Some languages support assignments as expressions, while others do not.

For instance, consider this JavaScript (see Appendix A of ES5 for the grammar rules).

{ x = y = f() + 2 }

In this case, the block is a statement (BlockStatement) and x = .. is also considered a "statement" (although it is really an Expression via Statement -> ExpressionStatement) while y = .. is an expression. Likewise, f() is an expression (technically, f is also an expression in JavaScript) and 2 is an expression and f() + 2 is an expression.

However, the following is invalid Pascal because Pascal's syntax does not support := (assignment) in an expression and an assignment is always a statement.

X := Y := F() + 2

Some languages also forbid general expressions as statements, which further throws off the notion that, in y = EXPR, it is correct to consider EXPR a valid statement. The following is invalid C#, but is dubiously valid in JavaScript and many other languages.

{ f() + 2; }

Answer 4

I would say that the "someFunction() + 2" is an expression being evaluated. Then I would say that "x = someFunction() + 2" is a statement, because most languages would generally evaluate the function's return plus two, and then assign that value to x.