The Nuances of Go: Go Program Structure

When declaring an entity within a function, it's scope is limited to that function. When declaring it outside of the function it is available to the contents of the package it is declared inside of. The case of the first letter also has the effect of allowing access within or across package boundaries. If the name begins with an uppercase character it is exported for other packages. Package names themselves are always lowercase, however.

Camel Casing

It's also important to know about the emphasis and standard within Go to use camel casing. Exceptions appear, however, when using acronyms such as HTML, XML, JSON, etc. When using such acronyms, it is routine to use names like HTMLthingy, XMLjunkHere, or rockStarJSON. Go is somewhat particular about these types of things, so it is extremely important to use the Go format standards. To help with this, there's also the gofmt function, but it's always good to just use the proper formatting while writing code anyway.

The Four Declarations

The four declarations are kind of like the four horsemen, except not really. Forget that analogy as it was already going nowhere. The four declarations include:

• var
• const
• type
• func

Here are some code samples of various declarations. The first code block uses func. In any new Go program there's always a main() function. The are declared like this:

func main() {
    // Stuff goes here.
}

Another example of a function declaration with parameters passed and expected results to return. The signature works like func, followed by the name of the function, then in parenthesis the parameter name followed by its type, for as many parameters as needed. Then after the parenthesis the type of the expected result of the function. Of course, if no result is expected to be returned, then there will be no type specified here. Just as if there were no parameters, the parenthesis would still be used but no parameters specified.

func doSomeComputation(someValue float64, anotherFloaty float64) float64 {
    return someValue * anotherFloat
}

Using var for the declaration of variables looks like this:

var f = 212.0
var result = 2 + 2
var celsius = (f - 32) * 5 / 9

Variable types can be defined in this way. Using the var keyword, followed by the variable name, then followed by its type, and then, if desired, assigned via the =, and then the specific value or function with a result to assign to the variable.

Without an assignment, the variable is simply created with the intent of assigning a value later. Such as these variable declarations.

var f float32
var result float64
var celsius float64
var someVariable int
var junk string

A shorter version of the creation and assignment of variables looks like this. It, however, can only be used within a function.

The actual use of this shorter version would look like this.

f := 212.0
randomStuff := rand.Float64()
wordContent := "This is some character string nonsense."

One can also do some other tricks with this shorter method of assignment. Let's say some variables are needed that will have integers assigned to them. You could set up the variables with values like this. I've noticed some get a bit confused about this and tuple assignments, this is not the same thing. This merely gives you three variables: k, l, and m and assigns the values 0, 1, and 2 to the respective variable.

k, l, m := 0, 1, 2

It's also important to note that a short declaration doesn't always declare all of the variables on the left-hand side. If already declared in the same lexical block, it then acts as an assignment to the variables instead.

Declaring a constant value with const would look like this:

const fahrenheitBoiling = 212.0

Pointers

A variable holds a particular value, whereas a pointer holds the address in memory of a particular value. Sometimes, it gets a bit difficult to mentally keep track of pointers, so let's look at a few examples. First I'll start with a plain old variable.

someVariable := "The original value of this variable."

Then I'll add another variable, which will point at the address and respective value of someVariable.

someVariable := "The original value of this variable."
anotherVariable := &someVariable
fmt.Println(*anotherVariable)

In this situation, what does the fmt.Println(*anotherVariable) function print out? The result would be "The original value of this variable." This is because anotherVariable is a variable that points to the value of someVariable. Which means anotherVariable doesn't take the resources that someVariable does, but merely points to it.

The reasoning here is that if you have a variable, for example, that stores a very large value in memory, one can merely point to that data with a pointer and the large value doesn't have to change its memory space. Imagine if that variable were a gigabyte of data and it had to be moved every time something needed to be referenced to it or calculated with it; with a pointer, we don't have to move it. It can be used in various ways without it getting shifted around.

Now I can take the pointer value, anotherVariable, and assign it a value, like this:

*anotherVariable = "42"

Then what happens? The pointer now is re-allocated and assigned the new value of "42." It no longer references the address space of the pointer and instead now has the value (42) assigned to the variable.

The Fancy new Function

There's also another function, a built-in one, that creates a variable. Using the expression new(type) will create an unnamed variable of type, initialize it, and return the address of the value of type. In pointer speak, it'd be *type. Here are some examples:

pinto := new(int)
fmt.Println(*pinto)

This code will create a variable of type int, then initialize the value to zero since it is an int. Let's say we want to actually give it a value after it's created, though. We could assign pinto like this:

*pinto = 42
fmt.Println(*p)

This will then print out the value 42, of type integer. Note, new is just sugar, a syntactical convenience feature really. They can be useful, however, and there are a few caveats that need to be noted. For example, check out this gotcha.

If I declare two new values using new.

firstThing := new(int)
secondThing := new(int)
fmt.Println(firstThing == secondThing)

In this case, the firstThing and secondThing don't actually match, because the comparison is on the address space, and they're different address spaces!

Alright, that's it for now, just a few important nuances around how variables, pointers, and Go features work.