Clean Go Code

Preface

The motivation behind writing this document, is to create a resource (and eventually a reference) for the Go community, which will help developers write cleaner code. This benefits every one of us. Whether we are writing code by ourselves, or writing code in larger teams. Establishing good paradigms for writing clean code and ensuring that this is available for everyone, will help prevent many meaningless hours on trying to understand and parse others (and our own) code.

We don’t read code, we decode it - Peter Seibel

The matter of the fact is, as Peter Seibel put it. We decode code and we honestly can't help encoding it, in some way, shape or form. This document, will be a precursor for us, to make sure that our encoding method is effective. We want our code to be usable, readable and maintainable.

This document will start with a simple and short introduction to the fundamentals behind writing clean code and will thereafter transition into concrete refactoring examples, more specific to Go. The aim of the document is to deliver the message of how easy it is to write clean code and how easy is it to write code, when it's clean.

A short word on `gofmt`

I would like to take a few sentences to make my stance on gofmt very clear. There are extremely many things that I disagree with, when it comes to gofmt. I prefer snake case over camel case, I quite like my constant variables to be upper case and I also have many opinions on bracket placement. That being said: gofmt is what enables us to have a common standard for writing Go code. All Go code, will look somewhat similar and it ensures that no Go code becomes too exoteric. I think that this is overall extremely positive. I appreciate immensely, that all Go programmers are somewhat restricted to write similar code, despite being very unhappy with some of the formatting rules. In my opinion, I value homogenous code over complete expressive freedom.

Introduction to Clean Code

Clean Code, is the pragmatic concept of ensuring readable and maintainable code. Clean Code establishes trust in the codebase and will steer developers away from introducing bugs. Clean Code will also establish much more stability in development speed, which typically will take a nose dive in the later stages of projects, due to higher risk of increasing bugs when introducing changes, as the codebase expands.

Test Driven Development

The core of creating clean code stems from creating good tests. Writing good tests helps create clean code, as it invites developers to think about the outcomes and test coverage of functions / functionality. It's easier to test a function that is only 4 lines, rather than a function, which is 40. In the same manner, a function which is 4 lines, is typically easier to understand than a function of 40 lines. Therefore, when using test driven development, the resulting code is much more likely to be of a cleaner nature.

The next important part of test driven development, which is very closely related to clean code, is the TDD cycle:

Write a test which fails
Make the test pass
Refactor code
Repeat

Step three of the cycle, ensures that we can refactor our code as we are writing it. The tests ensure that our refactor doesn't change the outcome of our functions and we can therefore, essentially, go crazy refactoring our code to be as clean as possible. As we go along, and our codebase expands, we will still have our tests, to make sure that our refactoring will not affect the outcome of our functions.

Naming

Comments

First things first: I want to address the topic of comments. Unnecessary comments are the biggest indicator of code smell. Comments are usually added into a code base, because something is so unclear, that it's deemed necessary to explain with a comment. Of course, this is not always the case. In Go, all according to gofmt all public variables and functions, should be annotated. I think this is absolutely fine, as this makes documentation easy. However, I therefore always want to distinguish between comments which enable auto-generated documentation and all other comments. Annotation comments, for documentation, should be written like documentation. They should be high level and concern the logical implementation as little as possible, other than on the highest abstraction level.

The reasoning behind this, is that there are other ways to explain code and ensure that code is being written comprehensibly and expressively. If the code is neither of the two, some people find it acceptable to replace this, with a comment explaining the logic. The matter of the fact is, that most people will not read comments, because it's very intrusive to the experience of reading code. However, let's start from the beginning. Bad comments:

// iterate over the range 0 to 9 
// and invoke the doSomething function
// for each iteration
for i := 0; i < 10; i++ {
  doSomething(i)
}

This comment, is what I call a tutorial comment. It's pretty common in tutorials, which explain the low level functionality of a language (or programming on a more general level). The matter of the fact is, that these comments are absolutely useless in production code. Hopefully, we aren't collaborating with other programmers, who don't understand that principles behind the language we have chosen to write in, or even worse, don't understand simple principles of programming. As programmers, we don't have to read the comment, we know this is happening, by reading the code. Hence the proverb:

"Document why, not how." - Venkat Subramaniam

Following this logic, we can now change our comment, to explain why we are iterating from the range zero to nine:

// instatiate 10 threads to handle upcoming work load
for i := 0; i < 10; i++ {
  doSomething(i)
}

Now we can understand why we are iterating and we can tell what we are doing, by reading the code. The worrying part about this comment is, that this probably should be necessary to express in prose. We can quite easily express this directly in our code instead:

for worker_id := 0; worker_id < 10; worker_id++ {
  instantiateThread(worker_id)
}

With just a few changes of our variable and function names, we have established explanations of our action, directly in our code. This is much clearer for the reader, because the reader won't have to read the comment and then map the prose comment to the code. Instead, they can read the code and immediately understand everything which is going on.

Of course, this example, was relatively easy. It's unfortunately not always this easy and writing clear and expressive code becomes increasingly difficult, together with code complexity. So let's have a look at some methods, which will help make this task easier.

Function Naming

We will start by discussing the naming of our functions. The general rule for function naming is really simple: The more specific the function, the more general the name. In other words, this means that we want to start with a very broad and short function name, such as Run or Parse, which describes the general functionality. Let's imagine that we are creating a configuration parser. Following this naming convention, our top level of abstraction might look something like the following:

func main() {
    configpath := flag.String("config-path", "", "configuration file path")
    flag.Parse()

    config, err := configuration.Parse(*configpath)
    
    ...
}

Our focus being on the naming of the Parse function. Despite this function name being very short and general, it's actually quite clear what this function attempts to achieve. When we go one layer deeper, our function naming will become slightly more specific:

func Parse(filepath string) (Config, error) {
    switch fileExtension(filepath) {
    case "json":
        return parseJSON(filepath)
    case "yaml":
        return parseYAML(filepath)
    case "toml":
        return parseTOML(filepath)
    default:
        return Config{}, ErrUnknownFileExtension
    }
}

More specific, but not much, just appropriately. It's still clear what the difference is between the parent function and the sub-functions, without being overly specific. This enables each sub-function to appear clear on it's own, whereas if we had named the parseJSON function json instead. This would not have been the case.

Notice that fileExtension is actually a little more specific. However, this is because the functionality of this function is, in fact, quite specific:

func fileExtension(filepath string) string {
    segemnts := strings.Split(filepath, ".")
    return segments[len(segments)-1]
}

This kind of logical progression in our function names, makes the code easier to follow and will make the code much easier to read. When we think about the opposite approach to function naming, it becomes even more clear why. If our highest level of abstraction becomes too specific, we will end up with a function name such as DetermineFileExtensionAndParseConfigurationFile. This is horrendously difficult to read and just adds confusion, more than anything else. We are trying to be too specific too quickly and therefore we end up being confusing, despite our intention of trying to be clear.

Variable Naming

Rather interestingly, the opposite is true for variables. Unlike functions, our variable naming should progress from more to less specific.

"You shouldn’t name your variables after their types for the same reason you wouldn’t name your pets 'dog' or 'cat'." - Dave Cheney

The reason why we want to become less and less specific with our variables, is the fact that it becomes clearer and clearer for the reader, what the variable represents, the smaller the scope of the variable is. In the example of the previous function fileExtension, the naming of the variable segments, could even be shortened to s, if we wanted to. The context of the variable is so clear, it is unnecessary to explain our code further, with longer variable names. Another good example of this, would be in nested for loops.

func PrintBrandsInList(brands []BeerBrand) {
    for _, b := range brands { 
        fmt.Println(b)
    }
}

The reason why this is true, is because of the scope of the variable, rather than the abstraction layer, which is the guideline we would use for our function naming. The smaller the scope of a variable, the less important the actual naming is. In the above example, the b variable scope is so short, that we don't need to spend brain power on remembering what it represents. However, because the scope brands is slightly larger, when reading the code, we will use more brain power on remembering what these represent. When expanding the variable scope in the function below, it becomes even more apparent:

func BeerBrandListToBeerList(beerBrands []BeerBrand) []Beer {
    var beerList []Beer
    for _, brand := range beerBrands {
        for _, beer := range brand {
            beerList = append(beerList, beer)
        }
    }
    return beerList
}

Now, let's imagine that we apply the opposite logic, to see what this looks like:

func BeerBrandListToBeerList(b []BeerBrand) []Beer {
    var bl []Beer
    for _, beerBrand := range b {
        for _, beerBrandBeerName := range beerBrand {
            bl = append(bl, beerBrandBeerName)
        }
    }
    return bl
}

Even though the function might still be readable, due to it's brevity, there is a strange off-putting feeling, when reading through the function. Should the scope of the variables or the logic of the function expand, this off-putting feel, becomes even worse and could potentially spiral into complete confusion. However, while on the topic of functions and their brevity, let's dive into the next topic of writing clean code.

Cleaning Functions

Function Length

In the words of Robert C. Martin:

"How small should a function be? Smaller than that!"

When writing clean code, our primary goal is to make our code easily digestible. The most effective way to do this, is to make our functions as small as possible. It's important to understand, that this is not necessarily to avoid code duplication. The more prominent reason for this is to heighten the code comprehension. Another way of explaining this, is to look at a function description:

fn GetItem:
    - parse json input for order id
    - get user from context
    - check user has appropriate role
    - get order from database

When using small functions (typically 5-8 lines in Go), we can create a function that reads almost as easily as our description:

var (
    NullItem = Item{}
    ErrInsufficientPrivliges = errors.New("user does not have sufficient priviliges")
)

func GetItem(ctx context.Context, json []bytes) (Item, error) {
    order, err := NewItemFromJSON(json)
    if err != nil {
        return NullItem, err
    }
    if !GetUserFromContext(ctx).IsAdmin() {
	      return NullItem, ErrInsufficientPrivliges
    }
    return db.GetItem(order.ItemID)
}

Using smaller functions also has a side-effect of eliminating another horrible habit of writing code: indentation hell. Indentation hell, typically occurs when a chain of if statements are clumsily inserted into a function. This makes the code very, very difficult to parse (for human beings) and should be eliminated whenever spotted. This is particularly common when working with interface{} and using type casting:

func GetItem(extension string) (Item, error) {
    if refIface, ok := db.ReferenceCache.Get(extension); ok {
        if ref, ok := refIface.(string); ok {
            if itemIface, ok := db.ItemCache.Get(ref); ok {
                if item, ok := itemIface.(Item); ok {
                    if item.Active {
                        return Item, nil
                    } else {
                      return EmptyItem, errors.New("no active item found in cache")
                    }
                } else {
                  return EmptyItem, errors.New("could not cast cache interface to Item")
                }
            } else {
              return EmptyItem, errors.New("extension was not found in cache reference")
            }
        } else {
          return EmptyItem, errors.New("could not cast cache reference interface to Item")
        }
    }
    return EmptyItem, errors.New("reference not found in cache")
}

Not only can this kind of code result in a really bad experience for other programmers, who will have to fight to understand the flow of the code. Should the logic in our if statements expand, it becomes exponentially more difficult to figure out which statement returns what. It is unfortunately not uncommon to find this kind of implementation in code. I have even bumped into examples of the beginning if statement of a corresponding else statement, was on another page of my monitor. Having to scroll up and down a page, while trying to figure out what a function does, is not ideal. Even though, we don't have to scroll on our page to see the corresponding if else statements in the above code sample, we are still scrolling with our eyes and maintaining state in our brain. Most programmers can quite easily contain this state for the function above, or worse examples. However, we have forced readers of our code, to use unnecessary brain power. This may result in reader fatigue, should we repeat this mistake throughout our code. Constantly having to parse code like the above, will make reading the code more and more difficult, which we of course, want to avoid.

So, how do we clean this function? Fortunately, it's actually quite simple. On our first iteration, we will try to ensure that we are returning an error as soon as we can. Instead of nested the if and else statements, we want to "push our code to the left". This is handled by returning from our function, as soon as we possibly can.

func GetItem(extension string) (Item, error) {
    refIface, ok := db.ReferenceCache.Get(extension)
    if !ok {
        return EmptyItem, errors.New("reference not found in cache")
    }

    if ref, ok := refIface.(string); ok {
        // return cast error on reference 
    }

    if itemIface, ok := db.ItemCache.Get(ref); ok {
        // return no item found in cache by reference
    }

    if item, ok := itemIface.(Item); ok {
        // return cast error on item interface
    }

    if !item.Active {
        // return no item active
    }

    return Item, nil
}

Once we are done with this, we can split up our function into smaller functions as mentioned previously. A good rule of thumb being: If the value, err := pattern is repeated more than once in a function, this is an indication that we can split the logic of our code into smaller functions.

func GetItem(extension string) (Item, error) {
    if ref, ok := getReference(extension) {
        return EmptyItem, ErrReferenceNotFound
    }
    return getItemByReference(ref)
}

func getReference(extension string) (string, bool) {
    refIface, ok := db.ReferenceCache.Get(extension)
    if !ok {
        return EmptyItem, false
    }
    return refIface.(string)
}

func getItemByReference(reference string) (Item, error) {
    item, ok := getItemFromCache(reference)
    if !item.Active || !ok {
        return EmptyItem, ErrItemNotFound
    }
    return Item, nil
}

func getItemFromCache(reference string) (Item, bool) {
    if itemIface, ok := db.ItemCache.Get(ref); ok {
        return EmptyItem, false
    }
    return itemIface.(Item), true
}

For production code, one should elaborate on the code even further, by returning errors instead of a bool values. This makes it much easier to understand where the error is originating from. However, as these are just example functions, the bool values will suffice for now. Examples of returning errors more explicitly will be explained in more detail later.

The resulting clean version of our function, has resulted in a lot more lines of code. However, the code is so much easier to read. It's layered in an onion-style fashion, where we can ignore code details that we aren't interested in and dive deeper into the functions that we wish to know the workings behind. When we do deep-dive into the lower level functionality, it will be extremely easy to comprehend, because we will only have to understand 3-5 lines in this case. This example illustrates, that we cannot score the cleanliness of our code from the line count of our functions. The first function iteration was much shorter. However, it was artificially short and very difficult to read. In most cases cleaning code will, to begin with, expand the already existing code base, in terms of lines of code. However, the benefit of readability is far preferred. If you are ever in doubt about this, think of how you feel about the following function, which does the same:

func GetItemIfActive(extension string) (Item, error) {
    if refIface,ok := db.ReferenceCache.Get(extension); ok {if ref,ok := refIface.(string); ok { if itemIface,ok := db.ItemCache.Get(ref); ok { if item,ok := itemIface.(Item); ok { if item.Active { return Item,nil }}}}} return EmptyItem, errors.New("reference not found in cache")
}

While we are on the topic. There are also a bunch of other side-effects that come along when writing in this style of code. Rather obviously, it makes our code much easier to test. It's much easier to get 100% code coverage on a function that is 4 lines (written by a sane person), than a function which is 400 lines. That's common sense.

Function Signatures

Creating good function naming structure, makes it easier to read and understand the intent of code. Making our functions shorter, helps with the understanding of the content of the function logic. The last part of cleaning our functions, will be to understand the context of the function input. With this, comes another easy to follow rule. Function signatures, should only contain one or two input parameters. On certain exceptional occasions, three can be acceptable, but this is where we should start considering a refactor. Much like the rule that our function should only be 5-8 lines long, this can seem quite extreme at first. However, I feel that this rule is more immediately demonstrably true.

As an example, take the following function from the RabbitMQ introduction tutorial, to their Go library:

q, err := ch.QueueDeclare(
  "hello", // name
  false,   // durable
  false,   // delete when unused
  false,   // exclusive
  false,   // no-wait
  nil,     // arguments
)

The function QueueDeclare takes six input parameters, which is quite extreme. The above code is somewhat possible to understand, because of the comments, but as mentioned earlier: Comments should be substituted with descriptive code. One good reason for this, is that there is nothing preventing us from invoking the QueueDeclare function without comments, making it look like this:

q, err := ch.QueueDeclare("hello", false, false, false, false, nil)

Now, without looking at the previous code, try to remember what the fourth and fifth false represent. It's impossible, and it's inevitable that we will forget at some point. This can lead to costly mistakes, and bugs that are difficult to correct. The mistakes might even occur through incorrect comments. Imagine labelling the wrong input parameter. Correcting this mistake, will be unbearably difficult to correct, especially when familiarity with the code has deteriorated over time or was low to begin with. Therefore, it is recommended to replace these input parameters, with an 'Options' struct instead:

type QueueOptions struct {
    Name string
    Durable bool
    DeleteOnExit bool
    Exclusive bool
    NoWait bool
    Arguments []interface{} 
}

q, err := ch.QueueDeclare(QueueOptions{
    Name: "hello",
    Durable: false,
    DeleteOnExit: false,
    Exclusive: false,
    NoWait: false,
    Arguments: nil,
})

This solves both the problem of omitting comments or accidentally labelling the variables incorrectly. Of course, we can still confuse properties with the wrong value, but in these cases, it will be much easier to determine where our mistakes lies within the code. The ordering of the properties also do not matter anymore and therefore incorrectly ordering the input values, is no longer a worry. The last added bonus of this technique, is that we can use our Option struct, to infer default values of our functions input parameters. When structures in Go are declared, all properties are initialised to their default value. This means, that our QueueDeclare option, can actually be invoked in the following way:

q, err := ch.QueueDeclare(QueueOptions{
    Name: "hello",
})

The rest of the values are by initialised to their default false values (except for Arguments, which, as an interface has a default value of nil). Not only are we much safer, we are more clear with our intentions and in this case, we could actually write less code. This is an all around win.

A last note on this, is that it's not always possible to change the function signatures. As in this case, we don't have control of our QueueDeclare function signature, since this is from the RabbitMQ library. It's not our code, we can't change it. However, we can wrap these functions, to suit our purposes:

type RMQChannel struct {
    channel *amqp.Channel
}

func (rmqch *RMQChannel) QueueDeclare(opts QueueOptions) (Queue, error) {
    return rmqch.channel.QueueDeclare(
        opts.Name,
        opts.Durable,
        opts.DeleteOnExit,
        opts.Exclusive,
        opts.NoWait,
        opts.Arguments, 
    )
}

Basically, we create a new structure RMQChannel which contains the amqp.Channel type, which has the QueueDeclare method. We then create our own version of this method, which essentially just calls the old version of the RabbitMQ library function. Our new method has all the advantages described before and we achieved this, without actually having access to changing any code in the RabbitMQ library.

We will use the idea of wrapping functions to introduce more clean and safe code later when discussing the interface{}.

RabbitMQ Go tutorial

Variable Scope

Now, let's go back one step, back to the idea of writing smaller functions. This has another nice side-effect, which we didn't cover in the previous chapter: Writing smaller function can typically eliminate using longer lasting mutable variables. Writing code with global variables, is a practice of the past, it doesn't belong in clean code. Now, why is that? Well, the problem with using global variables is that we make it very difficult for programmers to understand the current state of a variable. If a variable is global and mutable, then, by definition, it's value can be changed by any part of the codebase. At no point can you guarantee that this variable is going to be a specific value... This is a headache for everyone. This is yet another example of a trivial problem, which is exacerbate, when the codebase expands. Let's, look at a short example of how even larger scoped (not global) variables can cause problems.

Larger scoped variables, also introduce the issue of variable shadowing as shown int he code taken from an article named: Golang scope issue - A feature bug: Shadow Variables:

func doComplex() (string, error) {
    return "Success", nil
}

func main() {
    var val string
    num := 32

    switch num {
    case 16:
    // do nothing
    case 32:
        val, err := doComplex()
        if err != nil {
            panic(err)
        }
        if val == "" {
            // do something else
        }
    case 64:
        // do nothing
    }
    
    fmt.Println(val)
}

The problem with this code, from a quick skim, it seems like that the var val string value, should be printed out as: Success by the end of the main function. Unfortunately, this is not the case. The reason for this is, the line:

val, err := doComplex()

This declares a new variable val in the the switch case 32 scope and has nothing to do with the variable declared in the first line of main. Of course, it can be argued that the Go syntax is a little tricky, which I don't necessarily disagree with, but there is a much worse issue at hand. The declaration of var val string as a mutable largely scoped variable, is completely unnecessary. If we do a very simple refactor, we will no longer have this issue:

func getStringResult(num int) (string, error) {
    switch num {
    case 16:
    // do nothing
    case 32:
       return doComplex()
    case 64:
        // do nothing
    }
    return "" 
}

func main() {
    val, err := getStringResult(32)
    if err != nil {
        panic(err)
    }
    if val == "" {
        // do something else
    }
    fmt.Println(val)
}

After our refactor, val is no longer mutated and the scope has been reduced. Again, keep in mind that these functions are very simple. Once this kind of code style becomes a part of larger more complex systems, it can be impossible to figure out, why errors are happening. We don't want this to happen. Not only because we generally dislike errors happening in software, but it is also disrespectful to our colleagues, and ourselves, that we are potentially wasting each others live's, having to debug this type of code. Let's take responsibility ourselves, rather than blaming the variable declaration syntax in Go.

On a side not, if the // do something else part is another attempt to mutate the val variable. We should extract whatever logic in there as a function, as well as the previous part of it. This way, instead of prolonging the mutational scope of our variables, we can just return a new value:

func getVal(num int) (string, error) {
    val, err := getStringResult(32)
    if err != nil {
        return "", err
    }
    if val == "" {
        return NewValue() // pretend function
    }
}

func main() {
    val, err := getVal(32)
    if err != nil {
        panic(err)
    }
    fmt.Println(val)
}

Variable Declaration

Other than avoiding variable scope and mutability, we can also improve readability but keeping our variable declaration close to the logic. In C programming, it's common to see the following method for declaring variables:

func main() {
  var err error
  var items []Item
  var sender, receiver chan Item
  
  items = store.GetItems()
  sender = make(chan Item)
  receiver = make(chan Item)
  
  for _, item := range items {
    ...
  }
}

This suffers from the same symptom as described in variable scope. Even though that these variables might not actually be re-assigned at any point, this kind of style, will keep the readers on their toes, in all the wrong ways. Much like computer memory, our brain has a limited amount to allocate from. Having to keep track of which variables could be mutated and whether or not something will mutate these items, will only make it more difficult to get a good overview of what is happening in the code. Figuring out the eventually returned value, can be a nightmare. Therefore, to makes this easier for our readers, which could potentially be a future version of ourselves, it is good practice to declare variables as close to their usage as possible:

func main() {
  var sender chan Item
  sender = make(chan Item)
  
  go func() {
   	for {
    	select {
      case item := <- sender:
        // do something
      }
  	} 
  }()
}

However, we can do even better than this, by invoking the function directly on declaration. This makes it much clearer, that the function logic is associated with the declared variable, which is not as clear in the previous example.

func main() {
  sender := func() chan Item {
    channel := make(chan Item)
    go func() {
      for {
        select { ... }
      }
    }()
    return channel
  }
}

And coming full circle, we can move the anonymous function, to make it a named function instead:

func main() {
  sender := NewSenderChannel()
}

func NewSenderChannel() chan Item {
  channel := make(chan Item)
  go func() {
    for {
      select { ... }
    }
  }()
  return channel
}

It is still clear that we are declaring a variable and the logic, and the logic associated with the returned channel. Unlike, the first example. This makes it easier to traverse code and understand the responsibility of each variable.

Of course, this doesn't actually limit us from mutating our sender variable. There is nothing that we can do about this, as there is no way of declaring a const struct or static variables in Go. This means, that we will have to restrain ourselves from mutating this variable at a later point in the code.

NOTE: The keyword const does exist, but are limited for use on primitive types.

One way of getting around this, which at least will limit the mutability of a variable to a package level. Is to create a structure, with the variable as a private property. This private property is, thenceforth, only accessible through other methods of this wrapping structure. Expanding on our channel example, this would look something like the following:

type Sender struct {
  sender chan Item
}

func NewSender() *Sender {
  return &Sender{
    sender: NewSenderChannel(),
  }
}

func (s *Sender) Send(item Item) {
  s.sender <- item
}

We have now ensured, that the sender property of our Sender struct, is never mutated. At least not, from outside of the package. As of writing this document, this is the only way of creating publicly immutable non-primitive variables. It's a little verbose, but it's truly worth the effort, to ensure that we don't end up with strange bugs, that could be the outcome of mutating properties of our structure.

func main() {
  sender := NewSender()
  sender.Send(&Item{})
}

Looking at the example above, it's clear how this also simplifies the usage of our package. This way of hiding the implementation, is not only beneficial for the maintainers of the package, but also the users of the package. Now, when initialising and using the Sender structure, there is no concern of the implementation. This opens up, for a much looser architecture. Because our users aren't concerned with the implementation, we are free to change it at any point, since we have reduced the point of contact users of the package have. If we no longer wish to use a channel implementation in our package, we can easily change this, without breaking the usage of the Send method (as long as we adhere to it's current function signature).

NOTE: There is a fantastic explanation of how to handle the abstraction in client libraries, taken from the talk AWS re:Invent 2017: Embracing Change without Breaking the World (DEV319)

Clean Go

This section will describe some less generic aspects of writing clean Go code, but rather be discussing aspects that are very go specific. Like the previous section, there will still be a mix of generic and specific concepts being discussed, however, this section marks the start of the document, where the document changes from a generic description of clean code with Go examples, to Go specific descriptions, based on clean code principles.

Return Values

Returning Defined Errors

We will be started out nice an easy, by describing a cleaner way to return errors. Like discussed earlier, our main goals with writing clean code, is to ensure readability, testability and maintainability of the code base. This error returning method will improve all three aspects, with very little effort.

Let's consider the normal way to return a custom error. This is a hypothetical example taken from a thread-safe map implementation, we have named Store:

package smelly

func (store *Store) GetItem(id string) (Item, error) {
    store.mtx.Lock()
    defer store.mtx.Unlock()

    item, ok := store.items[id]
    if !ok {
        return Item{}, errors.New("item could not be found in the store") 
    }
    return item, nil
}

There is nothing inherently smelly about this function, in its isolation. We look into the items map of our Store struct, to see if we already have an item with this id. If we do, we return the item, if we don't, we return an error. Pretty standard. So, what is the issue with returning custom errors like this? Well, let's look at what happens, when we use this function, from another package:

func GetItemHandler(w http.ReponseWriter, r http.Request) {
    item, err := smelly.GetItem("123")
    if err != nil {
        if err.Error() == "item could not be found in the store" {
            http.Error(w, err.Error(), http.StatusNotFound)
	        return
        }
        http.Error(w, errr.Error(), http.StatusInternalServerError)
        return
    } 
    json.NewEncoder(w).Encode(item)
}

This is actually not too bad. However, there is one glaring problem with this. Errors in Go, are simply just an interface which implements a function (Error()) which returns a string. Therefore, we are now hardcoding the expected error code into our code base. This isn't too great. Mainly, because if the error message value changes, our code breaks (softly). Our code is too closely coupled, meaning that we would have to change our code in, possibly, many different places. Even worse would be, if a client used our package to write this code. Their software would inexplicably break all of a sudden after a package update, should we choose to change the message of the returning error. This is quite obviously something that we want to avoid. Fortunately, the fix is very simple.

package clean

var (
    NullItem = Item{}

    ErrItemNotFound = errors.New("item could not be found in the store") 
)

func (store *Store) GetItem(id string) (Item, error) {
    store.mtx.Lock()
    defer store.mtx.Unlock()

    item, ok := store.items[id]
    if !ok {
        return NullItem, ErrItemNotFound
    }
    return item, nil
}

With this simple change of making the error into a variable ErrItemNotFound, we ensure that anyone using this package can check against the variable, rather than the actual string that it returns:

func GetItemHandler(w http.ReponseWriter, r http.Request) {
    item, err := clean.GetItem("123")
    if err != nil {
        if err == clean.ErrItemNotFound {
           http.Error(w, err.Error(), http.StatusNotFound)
	        return
        }
        http.Error(w, err.Error(), http.StatusInternalServerError)
        return
    } 
    json.NewEncoder(w).Encode(item)
}

This feels much nicer and is also much safer. Some would even say that it's easier to read as well. In the case of a more verbose error message, it certainly would be preferable for a developer to simply read ErrItemNotFound rather than a novel on why a certain error has been returned.

This approach is not limited to errors and can be used for other returned values. As an example, we are also returning a NullItem instead of Item{} as we did before. There are many different scenarios in which it might be preferable to return a defined object, rather than initialising it on return.

Returning default Null values like the previous examples, can also be more safe, in certain cases. As an example, a user of our package could forget to check for errors and end up initialising a variable, pointing to an empty struct containing a default value of nil as one or more property values. When attempting to access this nil value later in the code, this could cause a panic in their code. However, when we return our custom default value instead, we can ensure that all values, which otherwise would default to nil, are initialised and thereby ensure that we do not cause panics in our users / clients software. This is also beneficial for ourselves, as if we wanted to achieve the same safety, without returning a default value, we would have to change our code, every place in which we return this type of empty value. However, with our default value approach, we now only have to change our code in a single place:

var NullItem = Item{
    itemMap: map[string]Item{},
}

NOTE: In many scenarios, invoking the panic will actually be preferable. To indicate that there is an error check missing.

NOTE: Every interface property in Go, has a default value of nil. This means that this is useful, for any struct, which has an interface property. This is also true for structs which contain channels, maps and slices, which could potentially also have a nil value.

Returning Dynamic Errors

There are certainly some scenarios, where returning an error variable might not actually be viable. In cases where customised errors' information is dynamic, to describe error events more specifically, we cannot define and return our static errors anymore. As an example:

func (store *Store) GetItem(id string) (Item, error) {
    store.mtx.Lock()
    defer store.mtx.Unlock()

    item, ok := store.items[id]
    if !ok {
        return NullItem, fmt.Errorf("Could not find item with ID: %s", id)
    }
    return item, nil
}

So, what to do? There is no well defined / standard method for handling and returning these kind of dynamic errors. My personal preference, is to return a new interface, with a bit of added functionality:

type ErrorDetails interface {
    Error() string
    Type() string
}

type errDetails struct {
    errtype error
    details string
}

func NewErrorDetails(err error, details ...interface{}) ErrorDetails {
    return &errDetails{
        errtype: err,
        details: details,
    }
}

func (err *errDetails) Error() string {
    return fmt.Sprintf("%v: %v", err.details)
}

func (err *errDetails) Type() error {
    return err.errtype
}

This new data structure still works as our standard error. We can still compare it to nil since it's an interface implementation and we can still call .Error() on it, so it won't break any already existing implementations. However, the advantage is that we can now check our error type as we could previously, despite our error now containing the dynamic details:

func (store *Store) GetItem(id string) (Item, error) {
    store.mtx.Lock()
    defer store.mtx.Unlock()

    item, ok := store.items[id]
    if !ok {
        return NullItem, fmt.Errorf("Could not find item with ID: %s", id)
    }
    return item, nil
}

And our http handler function can then be refactored to check for a specific error again:

func GetItemHandler(w http.ReponseWriter, r http.Request) {
    item, err := clean.GetItem("123")
    if err != nil {
        if err.Type() == clean.ErrItemNotFound {
            http.Error(w, err.Error(), http.StatusNotFound)
	        return
        }
        http.Error(w, err.Error(), http.StatusInternalServerError)
        return
    } 
    json.NewEncoder(w).Encode(item)
}

Nil Values

A controversial aspect of Go, is the addition of nil. This value corresponds to the value NULL in C and is essentially an uninitialised pointer. Previously, we traversed in explained the troubles nil can cause, but to sum up: Things break, when you try to access methods or properties of a nil value. In the mentioned section, it was recommended to try an minimise usage of returning a nil value. This way, users of our code, would be less prone to accidentally access nil values by a mistake.

There are other scenarios in which it is common to find nil values, which can cause some unnecessary pain. As an example, the incorrect initialisation of a struct can lead to the struct containing nil properties. If accessed, they will cause a panic. An example of this, can be seen below:

type App struct {
	Cache *KVCache
}

type KVCache struct {
  mtx sync.RWMutex
	store map[string]string
}

func (cache *KVCache) Add(key, value string) {
  cache.mtx.Lock()
  defer cache.mtx.Unlock()
  
	cache.store[key] = value
}

This code is absolutely fine. However, we are exposed by the fact that our App can be initialised incorrectly, without initialising our Cache property within. Should the following code be invoked, our application will panic:

	app := App{}
	app.Cache.Add("panic", "now")

The Cache property, has never been initialised and is therefore a nil pointer the Add method, is invoked. Running this code will result in a panic, with the following message:

panic: runtime error: invalid memory address or nil pointer dereference

Instead, we can turn our Cache property of our App structure into a private property and create a getter-like method, to access the Cache property of our App. This gives us more control of what we are returning and ensuring that we aren't returning a nil value.

type App struct {
    cache *KVCache
}

func (app *App) Cache() *KVCache {
	if app.cache == nil {
    app.cache = NewKVCache()
	}
	return app.cache
}

We now ensure that we will never experience returning a nil pointer, when trying to access the Cache property. Our code, which previously panicked, will now be refactored to the following:

app := App{}
app.Cache().Add("panic", "now")

The reason why this is preferable, is that we are ensuring that users of our package aren't worrying about the implementation and whether they are using our package in an unsafe manner. All they need to worry about is writing their own clean code.

NOTE: There are other methods to achieve a similar safe outcome, however, I think that this is the most straightforward method of doing this.

Pointers in Go

Pointers in go are rather a large topic. They are a very big part of working with the language, so much so, that it is essentially impossible to write go, without some knowledge of pointers and their workings in go. I will not go into detail, of the inner workings of Pointers in go in this article. Instead, we will focus on their quirks and how to handle them in go.

Pointers add complexity, however, as mentioned, it's almost impossible to avoid them when writing go. Therefore, it is important to understand how to use pointers, without adding unnecessary complexity and thereby keeping your codebase clean. Without restraining oneself, the incorrect use of pointers can introduce nasty side-effects, introducing bugs that are particularly difficult to debug. Of course, when sticking to the basic principles of writing clean code, introduced in the first part of this article, we limit our exposure of introducing this complexity, but pointers are a particular case, which can still undo all of our previous hard work, of making our code clean.

Pointer Mutability

I have already used the word mutability more than once in this article, as a negative. Mutability is obviously not a clear-cut bad thing and I am by no means an advocate for writing 100% pure functional programs. Mutability is a powerful tool, but we should really only ever use it, when it's necessary. Let's have a look at a code example illustrating why:

func (store *UserStore) Insert(user *User) error {
    if store.userExists(user.ID) {
        return ErrItemAlreaydExists
    }
    store.users[user.ID] = user
    return nil
}

func (store *UserStore) userExists(id int64) bool {
    _, ok := store.users[id]
    return ok
}

At first glance, this doesn't seem too bad. In fact, it might even seem like a rather simple insert function for a common list structure. We accept a pointer as input and if no other users with this id exist, then we insert the user pointer into our list. Now, we use this functionality in our public API for creating new users:

func CreateUser(w http.ResponseWriter, r *http.Request) {
    user, err := parseUserFromRequest(r)
    if err != nil {
        http.Error(w, err, http.StatusBadRequest)
        return
    }
    if err := insertUser(w, user); err != nil {
      http.Error(w, err, http.StatusInternalServerError)
      return
    }
}

func insertUser(w http.ResponseWriter, user User) error {
  	if err := store.Insert(user); err != nil {
        return err
    }
  	user.Password = ""
	  return json.NewEncoder(w).Encode(user)
}

Once again, at first glance everything looks fine. We parse the user from the received request and insert the user struct into our store. Once we have inserted our user into the store successfully, we then set the password to nothing, before returning the user as a JSON object to our client. This is all quite common practice, typically when returning a user object, where the password has been hashed, we don't want to return the hashed password.

However, imagine that we are using an in-memory store, based on a map, this code will produce some unexpected results. If we check in our user store, see that the change we made to the users password in the http handler function, also affected the object in our store. This is because the pointer address returned by parseUserFromRequest, is what we populated our store with, rather than an actual value. Therefore, when making changes to the dereferenced password value, we end up changing the value of the object we are pointing to in our store.

This is a great example of why both mutability and variable scope, can cause some serious issues and bugs, when used incorrectly. When passing pointers as an input parameter of a function, we are expanding the scope of our variable. Even more worrying, we are expanding the scope to an undefined level. We are almost expanding the scope of the variable to being a globally available variable. Depending on the variable scope of our store. As demonstrated by the above example, this can lead to disastrous bugs, which are particularly difficult to find and eradicate.

Fortunately, the fix for this bug, is rather simple:

func (store *UserStore) Insert(user User) error {
    if store.userExists(user.ID) {
        return ErrItemAlreaydExists
    }
    store.users[user.ID] = &user
    return nil
}

Instead of passing a pointer to a User struct, we are now passing in a copy of a User. We are still storing a pointer to our store, however, instead of storing the pointer from outside of the function, we are storing the pointer to the copied value, which scope is inside the function. This fixes the immediate problem, but might still cause issues further down the line, if we aren't careful.

func (store *UserStore) Get(id int64) (*User, error) {
    user, ok := store.users[id]
    if !ok {
        return EmptyUser, ErrUserNotFound
    }
    return store.users[id], nil
}

Again, a very standard very simple implementation of a getter function for our store. However, this is still bad. We are once again expanding the scope of our pointer, which may end up causing unexpected side-effects. When returning the actual pointer value, which we are storing in our user store, we are essentially giving other parts of our application the ability to change our store values. This is bad, because it's bound to ensure confusion. Our store should be the only entity enabled to make changes to the values stored there. The easiest fix available for this, is to either return a value of User rather than returning a pointer.

NOTE: Should our application use multiple threads, which is often the case. Passing pointers to the same memory location, can also potentially result in a race condition. In other words, we aren't only potentially corrupting our data, we could also cause a panic from a data race.

Please keep in mind, that there is intrinsically nothing wrong with returning pointers, however, the expanded scope and number of owners of the variables is the important aspect. This is what categorises our previous example a smelly operation. This is also why, that common Go constructors are also absolutely fine:

func AddName(user *User, name string) {
    user.Name = name
}

The reason why this is ok, is that the variable scope, which is defined by whomever invokes the functions, remains the same after the function returns. This combined with the face that the ownership of the variable remains unchanged (it stays solely with the function invoker), means that the pointer cannot be manipulated in an unexpected manner.

Closures are Function Pointers

So, before we go to the next topic of using interfaces in Go. I would like to introduce the commonly overseen alternative, which is what C programmers know as 'function pointers' and most other programmers refer to as 'closures'. Closure are quite simple. They are an input parameter for a function, which act like any other parameter, except for the fact that they are a function. In Javascript, it is very common to use closures as callbacks, which is typically used in scenarios where upon we want to invoke a function after an asynchronous operation has finished. In Go, we don't really have this issue, or at the very least, we have other, much nicer, ways of solving this issue. Instead, in Go, we can use closures to solve a different hurdle: The lack of generics.

Now, don't get too excited. We aren't going to substitute the lack of generics. We are simply going to solve a subset of the lack of generics with the use of closures. Consider the following function signature:

func something(closure func(float64) float64) float64 { ... }

This function takes another function as input and will return a float64. The input function, will take a float64 as input, and will also return a float64. This pattern can be particularly useful, for creating a loosely coupled architecture, making it easier to to add functionality, without affecting other parts of the code. An example use case of this, could be for a struct containing data, which we want to manipulate in some form. Through this structures Do() method, we can perform operations on this data. If we know the operation ahead of time, we can approach problem this by placing the logic for handling the different operations, directly in our Do() method:

func (datastore *Datastore) Do(operation Operation, data []byte) error {
  switch(operation) {
  case COMPARE:
    return datastore.compare(data)
  case CONCAT:
    return datastore.add(data)
  default:
    return ErrUnknownOperation
  }
}

As we can imagine, this function will perform a predetermined operation on the data contained in the Datastore struct. However, we can also imagine, that at some point we would want to add more operations. Over a longer period of time, this might end up being quite a lot of different operations, making our Do method bloated and possibly even hard to maintain. It might also be an issue for people wanting to use our Datastore object, who don't have access to edit our package code. Keeping in mind, that there is no way of extending structure methods as there is in most OOP languages. This could also become an issue for developers wanting to use our package.

So instead, let's try a different approach, using closures instead:

func (datastore *Datastore) Do(operation func(data []byte, data []byte) ([]byte, error), data []byte) error {
  result, err := operation(datastore.data, data)
  if err != nil {
    return err
  }
  datastore.data = result
  return nil
}

func concat(a []byte, b []byte) ([]byte, error) {
  ...
}

func main() {
  ...
  datastore.Do(concat, data)
  ...
}

However, other than this being a very messy function signature, we also have another issue with this. This function isn't particularly generic. What happens, if we find out that we actually want the concat function needs to be able to take multiple byte arrays as input? Or if want to add some completely new functionality, that may also need more or less input values than (data []byte, data []byte) ?

One way to solve this issue, is to change our concat function. In the example below, I have changed it to only take a single byte array as input argument, but it could just as well have been the opposite case.

func concat(data []byte) func(data []byte) ([]byte, error) {
  return func(concatting []byte) ([]byte, error) {
    return append(data, concatting), nil
  }
}

func (datastore *Datastore) Do(operation func(data []byte) ([]byte, error)) error {
  result, err := operation(datastore.data)
  if err != nil {
    return err
  }
  datastore.data = result
  return nil
}

func main() {
  ...
  datastore.Do(compare(data))
  ...
}

Notice how we have added some of the clutter from the Do method signature. The way that we have accomplished this, is by having our concat function return a function. Within the returned function, we are storing the input values originally passed in to our concat function. The returned function can therefore now take a single input parameter, and within our function logic, we will append it, with our original input value. As a newly introduced concept, this is quite strange, however, getting used to having this as an option can indeed help loosen up program coupling and help get rid of bloated functions.

In the next section, we will talk about interfaces, but let's take a short moment to talk about the difference between interfaces and closures. The problems that interfaces solve, definitely overlap with the problems solved by closures. The implementation of interfaces in Go makes the distinction of when to use one or the other, somewhat difficult at times. Usually, whether an interface or a closure is used, is not really of importance and whichever solves the problem in the simplest manner, is the right choice. Typically, closures will be simpler to implement, if the operation is simple by nature. However, as soon as the logic contained within a closure becomes complex, one should strongly consider using an interface instead.

Dave Cheney has an excellent write up on this topic, and a talk on the same topic:

Jon Bodner also has a talk about this topic

https://www.youtube.com/watch?v=5IKcPMJXkKs

Interfaces in Go

In general, the go method for handling interface's is quite different from other languages. Interfaces aren't explicitly implemented, like they would be in Java or C#, but are implicitly implemented if they fulfill the contract of the interface. As an example, this means that any struct which has an Error() method, implements / fulfills the Error interface and can be returned as an error. This has it's advantages, as it makes Go feel more fast-paced and dynamic, as interface implementation is extremely easy. There are obviously also disadvantages with this approach to implementing interfaces. As the interface implementation is no longer explicit, it can be difficult to see which interfaces are implemented by a struct. Therefore, the most common way of defining interfaces, is by writing interfaces with as few methods a possible. This way, it will be easier to understand whether or not a struct fulfills the contract of an interface.

There are other ways of keeping track of whether your structs are fulfilling the interface contract. One method, is to create constructors, which return an interface, rather than the concrete type:

type Writer interface {
	Write(p []byte) (n int, err error)
}

type NullWriter struct {}

func (writer *NullWriter) Write(data []byte) (n int, err error) {
    // do nothing
    return len(data), nil
}

func NewNullWriter() io.Writer {
    return &NullWriter{}
}

The above function ensures, that the NullWriter struct implements the Writer interface. If we were to delete the Write method for the NullWriter we would get a compilation error, were we to try and build the solution. This is a good way of ensuring our code behaves in the way that we expect and that we can use the compiler as a safety net to ensure that we aren't producing invalid code.

There is another way of trying to be more explicit about which interfaces a given struct implements. However, this method achieves the opposite of what we wish to achieve. The method being, using embedded interfaces, as a struct property.

"Wait what?" - Presumably most people

So, let's rewind a little, before we dive deep into the forbidden forest of smelly Go. In Go, we can use embedded structs, as a type of inheritance in our struct definitions. This is really nice as we can decouple our code, by defining reusable structs.

type Metadata struct {
    CreatedBy types.User
}

type Document struct {
    *Metadata
    Title string
    Body string
}

type AudioFile struct {
    *Metadata
    Title string
    Body string
}

Above, we are defining a Metadata object, which will provide us with property fields that we are likely to use on many different struct types. The neat thing about using the embedded struct, rather than explicitly defining the properties directly in our struct, is that it has decoupled the Metadata fields. Should be choose to update our Metadata object, we can change it in a single place. As mentioned earlier, we want to ensure that a change one place in our code, doesn't break other parts of our code. Keeping these properties centralised, will keep it clear to users that a structures with embedded Metadata have the same properties. Much like, structures that fulfill interfaces, have the same methods.

Now, let's look at an example of how we can use a constructor, to further prevent breaking our code, when making changes to our Metadata struct:

func NewMetadata(user types.User) Metadata {
    return &Metadata{
        CreatedBy: user,
    }
}

func NewDocument(title string, body string) Document {
    return Document{
        Metadata: NewMetadata(),
        Title: title,
        Body: body,
    }
}

At a later point in time, we find out, that we would also like a CreatedAt field on our Metadata object. This is now easily achievable, by simply updating our NewMetadata constructor:

func NewMetadata(user types.User) Metadata {
    return &Metadata{
        CreatedBy: user,
        CreatedAt: time.Now(),
    }
}

Now, both our Document and AudioFile structures are updated, to also populate these fields on construction. This is the core principle behind decoupling and an excellent example of ensuring maintainability of code. We can also add new methods, without breaking our code:

type Metadata struct {
    CreatedBy types.User
    CreatedAt time.Time
    UpdatedBy types.User
    UpdatedAt time.Time
}

func (metadata *Metadata) AddUpdateInfo(user types.User) {
    metadata.UpdatedBy = user
    metadata.UpdatedAt = time.Now()
}

Again, without breaking the rest of our code base, we are implementing new functionality to our already existing structures. This kind of programming, makes implementing new features very quick and very painless, which is exactly what we are trying to achieve by making our code clean.

Now, I am sorry to break this streak of happiness, because now we return to the smelly forbidden forest of Go. Let's get back to our interfaces and how to show explicitly which interfaces are being implemented by a structure. Instead of embedding a struct, we can embed an interface:

type NullWriter struct {
    Writer
}

func NewNullWriter() io.Writer {
    return &NullWriter{}
}

The above code compiles. The first time I saw this, I couldn't believe that this was actually compiling. Technically, we are implementing the interface of Writer, because we are embedding the interface and "inheriting" the functions which are associated with this interface. Some see this as a clear way of showing that our NullWriter is implementing the Writer interface. However, we have to be careful using this technique, as we can no longer rely on the compiler to save us:

func main() {
    w := NewNullWriter()

    w.Write([]byte{1, 2, 3})
}

As mentioned before, the above code will compile. The NewNullWriter returns a Writer and everything is honky-dori, according to the compiler, because NullWriter fulfills the contract of io.Writer, via. the embedded interface. However, running the code above will result in the following:

panic: runtime error: invalid memory address or nil pointer dereference

The explanation being, that an interface method in Go, is essentially a function pointer. In this case, since we are pointing the function of an interface, rather than an actual method implementation, we are trying to invoke a function, which is in actuality a nil pointer. Oops! Personally, I think that this is a massive oversight in the Go compiler. This code should not compile... but while this is being fixed (if it ever will be), let's just promise each other, never to implement code in this way. In an attempt to be more clear with our implementation, we have ended up shooting ourselves in the foot and bypassing compiler checks.

Some people argue that using embedded interfaces, is a good way of creating a mock structure, for testing a subset of interface methods. Essentially, by using an embedded interface, you won't have to implement all of the methods of an interface, but instead only implement the few methods that you wish to be tested. Within testing / mocking, I can see the argument, but I am still not a fan of this approach.

Let's quickly get back to clean code and quickly get back to using interfaces the proper way in Go. Let's talk about using interfaces as function parameters and return values. The most common proverb for interface usage with functions in Go is:

"Be conservative in what you do, be liberal in what you accept from others" - Jon Postel

FUN FACT: This proverb originally has nothing to do with Go, but is actually taken from an early specification of the TCP networking protocol.

In other words, you should write functions that accept an interface and return a concrete type. This is generally good practice, and becomes super beneficial when doing tests with mocking. As an example, we can create a function which takes a writer interface as input and invokes the Write method of that interface.

type Pipe struct {
    writer io.Writer
    buffer bytes.Buffer
}

func NewPipe(w io.Writer) *Pipe {
    return &Pipe{
        writer: w,
    }
} 

func (pipe *Pipe) Save() error {
    if _, err := pipe.writer.Write(pipe.FlushBuffer()); err != nil {
        return err
    }
    return nil
}

Let's assume that we are writing to a file when our application is running, but we don't want to write to a new file for all tests which invokes this function. Therefore, we can implement a new mock type, which will basically do nothing. Essentially, this is just basic dependency injection and mocking, but the point is that it is extremely easy to use in go:

type NullWriter struct {}

func (w *NullWriter) Write(data []byte) (int, error) {
    return len(data), nil
}

func TestFn(t *testing.T) {
    ...
    pipe := NewPipe(NullWriter{})
    ...
}

NOTE: there is actually already a null writer implementation built into the ioutil package named Discard

When constructing our Pipe struct with the NullWriter (rather than a different writer), when invoking our Save function, nothing will happen. The only thing we had to do, was add 4 lines of code. This is why in idiomatic go, it is encouraged to make interface types as small as possible, to make implement a pattern like this as easy as possible. However, this implementation of interfaces, also comes with a huge downside.

The empty `interface{}`

Unlike other languages, go does not have an implementation for generics. There have been many implementation proposals, but all have been deemed dissatisfactory by the Go language team. Unfortunately, without generics, developers are trying to find creative ways around this issue, very often using the empty interface{}. The next section, will describe why these, often too creative, implementations should be considered bad practice and unclean code. There will also be good examples of usage of the empty interface{} and how to avoid some pitfalls of writing code with the empty interface{}.

But first and foremost. What drives developers to use the empty interface{}? Well, as I said in the previously, the way that Go determines whether a concrete type implements an interface, is by checking whether it implements the methods of a specific interface. So what happens, if our interface implement no methods at all?

type EmptyInterface interface {}

The above being equivalent to the built-in type interface{}. The result of this interface type is that any type is accepted. Meaning, that we can write functions in which any type is accepted. This is super useful for certain kind of functions, such as when creating a printer function. This is how it's possible to give any type to the Println function from the fmt package:

func Println(v ...interface{}) {
    ...
}

In this case, we aren't only accepting a single interface{} but rather, a slice of types. These types can be of any type and can even be of different types, as long as they implement the empty interface{}, which we are certain that any type will. This is a super common pattern when handling string conversation (both from and to string). The reason being, this is the only way in Go to implement generic methods. Good examples of this, come from the json standard library package:

func InsertItemHandler(w http.ResponseWriter, r *http.Request) {
	var item Item
	if err := json.NewDecoder(r.Body).Decode(&item); err != nil {
		http.Error(w, err.Error(), http.StatusBadRequest)
		return
	}

	if err := db.InsertItem(item); err != nil {
		http.Error(w, err.Error(), http.StatusInternalServerError)
		return
	}
	w.WriteHeader(http.StatsOK)
}

All the less elegant code, is contained within the Decode function. Developers using this functionality, therefore, won't have to worry about reflection or casting of types. We just have to worry about providing a pointer to a concrete type. This is good, because the Decode() function is, technically, returning a concrete type. We are passing in our Item value, which will be populated from body of the http request and we won't have to deal with the potential risks of handling the interface{} value.

However, even when using the empty interface{} with good practices, we still have some issues. If we pass a JSON string that has nothing to do with our Item type, but is still valid son, we still won't receive an error. Our item variable will just be left with the default values. So, while we don't have to worry about reflection and casting errors, we will still have to make sure that the message sent from our client is a valid Item type. However, as of writing this document, there is no simple / good way to implement these type of generic decoders, without using the empty interface{} type.

The problem with this, is that we are leaning towards using Go (a statically typed language) as a dynamically typed language. This becomes even clearer, when looking at poor implementations of the interface{} type. The most common example of this, comes from developers trying to implement a generic store / list of some sort. Let's look at an example, trying to implement a generic HashMap package, which can store any type, using the interface{}.

type HashMap struct {
    store map[string]interface{}
}

func (hashmap *HashMap) Insert(key string, value interface{}) {
    hashmap.store[key] = value
}

func (hashmap *HashMap) Get(id string) (interface{}, error) {
    value, ok := hashmap.store[key]
    if !ok {
        return nil, ErrKeyNotFoundInHashMap
    }
    return value
}

NOTE: I have omitted thread-safety from the example for simplicity

Please keep in mind that the implementation pattern used above, is used in quite a lot of Go packages. It is even used in the standard library sync package, for the sync.Map type. So, what is the big problem with this implementation? Well, let's have a look at an example of using this package.

func SomeFunction(id string) (Item, error) {
    itemIface, err := hashmap.Get(id)
    if err != nil {
        return EmptyItem, err
    }
    item, ok := itemIface.(Item)
    if !ok {
        return EmptyItem, ErrCastingItem
    }
    return item, nil
}

On first glance, this looks fine. However, like mentioned previously. However, we will start getting into trouble, should we add different types in our store, which as of now, is not prevented. There is nothing limiting us from adding something other than the Item type. So what happens when someone starts adding other types into our HashMap? Our function now might return an error. This might even be a small change like someone else in the code base wanting to store a pointer *Item instead of an Item. Worst of all, this might not even be caught by our tests. Depending on the complexity of the system, this might introduce some bugs particularly difficult to debug.

This type of code, should never reach production. The matter of the fact is, that Go does not support generics as of now and as Go programmers, we should accept this. If we want to use generics, then we should use a different language which does support generics, rather than trying hack our way out of this.

So, how do we prevent this code from reaching production? The simples solution for our problem, is basically to just write the functions with concrete types, instead of using interface{} values. Of course, this is not always the best approach, as there might be some functionality within the package which is not trivial to implement ourselves. Therefore, it might be a better approach to create wrappers, which expose the functionality we need, but still ensure type safety:

type ItemCache struct {
  kv tinykv.KV
} 

func (cache *ItemCache) Get(id string) (Item, error) {
  value, ok := cache.kv.Get(id)
  if !ok {
    return EmptyItem, ErrItemNotFound
  }
  return interfaceToItem(value)
}

func interfaceToItem(v interface{}) (Item, error) {
  item, ok := v.(Item)
  if !ok {
    return EmptyItem, ErrCouldNotCastItem
  }
  return item, nil
}

func (cache *ItemCache) Put(id string, item Item) error {
  return cache.kv.Put(id, item)
}

NOTE: Implementations of other functionalities of the tinykv.KV cache has been left out for the purpose of brevity.

Creating the wrapper above, will now ensure that we are using the actual types and that we are no longer passing in interface{} types. It is therefore no longer possible to accidentally populate our store with a wrong value type and we have contained our casting of types, as much as possible. This is a very straight forward way of solving our issue, though somewhat manual.

Summary

First of all, thank you for making it all the way through the article. I hope that it has given some insight into what clean code is, as well as how it will help ensure maintainability, readability and stability in your code base. To sum up all the topics covered:

Functions - Naming of functions should become more specific, the smaller the scope of the function. Ensure that all functions are single purpose. A good measure, being to limit your function length to 5-8 lines and only takes 2-3 input arguments.

Variables - Naming of variables should become less specific the smaller the scope, and keep the scope of your variables to a minimum. Also, keep the mutability of your variables to a minimum and be more and more aware of this as their scope grows.

Return Values - Concrete types should be returned whenever they can. Make it as hard as possible for users of your package to create mistakes and as easy for them to understand the values returned by your functions

Pointers - Use pointers with caution and limit scope and mutability to an absolute minimum. Garbage collection only assists with memory management, it does not assist with all the other complexities associated with pointers.

Interfaces - Use interfaces as much as possible to loosen the coupling of your code. Contain any code using the empty interface{} as much as possible and prevent it being exposed.

Of course, what is considered clean code is particularly subjective and I don't think that will ever change. However, much like my statement concerning gofmt, I think it's more important to find a common standard, rather than a standard that everyone agrees with 100%. It's also important to understand that fanaticism is never the goal. A codebase will most likely never be 100% 'clean', in the same way as your office desk isn't either. There is room for stepping outside the rules and boundaries established in this article. However, remember that the most important aspect of writing clean code, is helping one another. We help our support engineers, by ensuring stability in software and easy debugging. We help our fellow developers by ensuring our code is readable and easily digestible. We help everyone involved in the project by establishing a flexible code base, in which we can quickly introduce new features without breaking our current platform. We move quickly by going slowly and thenceforth, everyone is satisfied.

I therefore hope, that you will join the discussion to help what we, the Go community, define as clean code. Let's establish a common ground, so that we improve software. Not only for ourselves, but the sake of everyone.

hhy5277 / clean-go-article Goto Github PK

clean-go-article's Introduction