This repo highlights the list of software engineering guidelines in general. Most of these are industry-wide conventions, thus using them will ensure that your code is easily readable by people who are not you.
Just keep in mind that this post isn't about how you should indent your code, but it's more of a guidlines on how to write clean code that are easy to manage. With that said, if you are leading a team or even a single contributor developer who wanted to become better, having a set of coding guidelines is a great start to make that happen.
Also, feel free to add your own tips through pull requests.
int x = 32;
object o = x;
Boxing is the process of wrapping a value type such as a primitive or struct inside an object that lives on the heap. Unboxing is gettting the orginal value back out again.
đź‘Ť Map to matching primitive or struct variable instead:
int x = 32;
int y = x;
int statusCode;
if (condition)
{
statusCode = 1;
}
else
{
statusCode = 2;
}
đź‘Ť Do use ternary conditional operator (?:) instead:
int statusCode = condition ? 1 : 2;
The preceding code is much cleaner, easier to read and understand. On top of that, it's more concise.
public string GetJobStatus(Job job)
{
return job.IsRunning ? "Running" : job.HasErrors ? "Failed" : "Succeeded";
}
đź‘Ť Use another line to express the nested operation as a separate statement for readability:
if (job.IsRunning)
{
return "Running";
}
return job.HasErrors ? "Failed" : "Succeeded";
Nesting ternary operators results in the kind of code that may seem clear as day when you write it, but six months later could leave maintainers (or future you) scratching their heads.
if (something != null)
{
if (something.Other != null)
{
return something.Other.Whatever;
}
}
đź‘Ť Do use null conditional (?.) operator instead:
return something?.Other?.Whatever;
The preceding code is also much cleaner and concise.
if (something != null)
{
if (something.other != null)
{
return something.other.whatever;
}
else
{
return string.empty;
}
}
else
{
return string.empty;
}
đź‘Ť Do use null coalescing (??) operator instead:
return something?.other?.whatever ?? string.empty;
int? number = null;
var n = number.HasValue ? number : 0;
đź‘Ť Do use null coalescing (??) operator as well:
var n = number ?? 0;
đź‘Ť or alternatively, you could do:
var n = number.GetValueOrDefault();
Try to avoid using the equality operator (==) or HasValue for nullable variable check like in the following:
int? number = null;
if (number == null)
{
//do something
}
if (!number.HasValue)
{
//do something
}
đź‘Ť While the preceding code is fine, we can still improve that by using the is keyword like in the following:
int? number = null;
if (number is null)
{
//do something
}
Avoid code without braces ({}) for single conditional if statement, for and foreach loops like in the following:
if (condition) action;
Without the braces, it is too easy to accidentally add a second line thinking it is included in the if, when it isn’t.
đź‘Ť Always use braces instead:
if (condition) { action; }
//or better
if (condition)
{
action;
}
if (condition)
{
//do something
}
else if (condition)
{
//do something
}
else if (condition)
{
//do something
}
else (condition)
{
//do something else
}
đź‘Ť Do use switch statements instead:
switch (condition)
{
case 1:
//do something
break;
case 2:
//do something
break;
case 3:
//do something
break;
default:
//do something else
break;
}
đź‘Ť But prefer switch expressions over switch statements where possible like in the following:
condition switch
{
1 => //do something;
2 => //do something;
3 => //do something;
_ => //do something else;
}
The preceding code is more concise yet, still easy to read and understand. (Note, only available in C# 8 or later versions) đź’ˇ Exceptions - There are cases that if-else statements would make more sense than using switch statements. One such example could be, if the condition involves different objects and complex conditions, though if the conditional logic here boils down to checking whether an object matches a certain shape in terms of property values, you might want to explore recursive pattern matching.
Do use the using statement when working with objects that eat resources or implements IDisposable interface:
using (MemoryStream stream = new MemoryStream())
{
// do something
}
đź‘Ť Or prefer to use the new using declaration introduced in C# 8 like in the following:
using var stream = new MemoryStream();
// do something
The preceding code reduces the number of curly braces in your method, but it can still be seen easily where a resource is disposed. For more information, see: "pattern-based using" and "using declarations".
âš Do not use either of the above if the IDisposable is the return value of your method:
Stream GetABrokenFileStream()
{
// this is wrong! The stream will be disposed when you exit the method
using var fileStream = File.OpenRead("path to my file");
return fileStream;
}
string space = "Vianne";
string greetings = "Hello " + name + "!";
đź‘Ť Use string.Format() method instead:
string name = "Vynn";
string greetings = string.Format("Hello {0}!", name);
đź‘Ť Or prefer using string interpolation ($) instead where possible:
string name = "Vjor";
string greeting = $"Hello, {name}!;
The preceding code is much more concise and readable compared to other approaches.
var date = DateTime.Now;
string greetings = string.Format("Today is {0}, the time is {1:HH:mm} now.", date.DayOfWeek, date);
đź‘Ť Use string interpolation instead:
var date = DateTime.Now;
string greetings = $"Today is {date.DayOfWeek}, the time is {date:HH:mm} now.");
The preceding code is much easier to understand and concise. However, there are certain cases that using the string.Format() would makes more sense. For example, when dealing with complex formatting and data manipulation. So, use your judgement when to apply them in situations.
List<Repository.DataAccessLayer.Whatever> listOfBlah = _repo.DataAccessLayer.GetWhatever();
đź‘Ť Use the var keyword instead:
var listOfBlah = _repo.DataAccessLayer.GetWhatever();
đź‘Ť Same goes for other local variables:
var students = new List<Students>();
var memoryStream = new MemoryStream();
var dateUntilProgramExpiry = DateTime.Now;
public string Greeter(string name)
{
return $"Hello {name}!";
}
đź‘Ť Do use Expression-bodied (=>) implementation instead:
public string Greeter(string name) => $"Hello {name}!";
The preceding code is more concise while maintaining readability.
Person person = new Person();
person.FirstName = "Vianne";
person.LastName = "Durano";
đź‘Ť Do use object and collection initializers instead:
var person = new Person
{
FirstName = "Vianne",
LastName = "Durano"
};
The preceding code is more natural to read and the intent is clear because the properties are defined within braces.
public Person GetName()
{
var person = new Person
{
FirstName = "Vincent",
LastName = "Durano"
};
return person;
}
đź‘Ť Do use Tuples instead where possible:
public (string FirstName, string LastName) GetName()
{
return ("Vincent", "Durano");
}
The preceding code is more convenient for accessing objects and manipulating the data set. Tuples replaces the need to create a new class whose sole purpose is to carry around data.
Try to create an Extention Methods to perform common tasks such as conversion, validation, formatting, parsing, transformation, you name it. So, instead of doing the following:
string dateString = "40/1001/2021";
var isDateValid = DateTime.TryParse(dateString, out var date);
The preceding code is perfectly fine and should handle the conversion safely. However, the code is bit lengthy just to do basic conversion. Imagine you have tons of the same code conversion cluttering within the different areas in your project. Your code could turn into a mess or potentially causes you alot of development time overtime.
đź‘Ť To prevent that, you should consider creating a helper/utility functions to do common tasks that can be reused across projects. For example, the preceding code can now be converted to following extension:
public static class DateExtensions
{
public static DateTime ToDateTime(this string value)
=> DateTime.TryParse(value, out var result) ? result : default;
}
and you will be able to use the extension method like in the following anywhere in your code:
var date = "40/1001/2021".ToDateTime();
The preceding code makes your code concise, easy to understand and provides convenience.
The preceding code removes alot of noise in your code when injecting dependencies as you don't need to write private readonly declarations which can make your code cleaner.
In situations where you want to expose one of the fields to be public, you can define and set it in the constructor as what you would normally do. Otherwise, the arguments are marked as private fields.
String firstName;
Int32 orderCount;
Boolean isCompleted;
đź‘Ť Do use built-in primitive data types instead:
string firstName;
int orderCount;
bool isCompleted;
The preceding code is consistent with the Microsoft’s .NET Framework and more natural to read.
private readonly PersonManager _pm;
The main reason for this is that it can cause confusion and inconsistency when you have class that might represents the same thing like in the following:
private readonly ProductManager _pm;
đź‘Ť Instead, do choose clarity ver brevity like in the following:
private readonly PersonManager _personManager;
private readonly ProductManager _productManager;
The preceding code provides more clarity as it clearly suggests what the object is about.
namespace ProjectName.App.Web
namespace ProjectName.Services.Common
namespace ProjectName.Services.Api.Payment
namespace ProjectName.Services.Api.Ordering
namespace ProjectName.Services.Worker.Ordering
Generally namespaces should reflect the folder hierarchy within a project. The preceding code suggest good organization of your code within the project, allowing you to navigate between layers easily.
public class Person
{
//some code
}
public class BusinessLocation
{
//some code
}
public class DocumentCollection
{
//some code
}
This enables you to easily determine if an object holds a single item value or collection. Imagine, if you have a List vs List. It's just odd to put plural form names in a List or Collection.
Do use nouns or adjective phrases for Property names as well. When naming boolean properties or variables, you may add the prefix "can", "is", "has", etc. just like in the following:
public bool IsActive { get; set; }
public bool CanDelete { get; set; }
//variables
bool hasActiveSessions = false;
bool doesItemExist = true;
Adding those suffixes will provide more value to the caller.
public class ClassName
{
const int MaxPageSize = 100;
public string PropertyName { get; set; }
public void MethodName()
{
//do something
}
}
This is so that our code is consistent with the Microsoft .NET Framework.
public void MethodName(CreatePersonRequestDto requestDto)
{
var firstName = requestDto.FirstName;
}
This is so that our code are consistent with the Microsoft .NET Framework.
int daysUntilProgramExpiry;
public Person GetPersonProfileById(long personId)
{
//do something
}
This makes your code easier to read and understand without having you to write (or atleast minimizes) comments of what the code does.
public async Task<List<Person>> GetPersonProfileByIdAsync(long personId)
{
//do something
}
This enable developers to easily identify synchronous vs asynchronous methods by just looking at the method name itself.
public interface IPersonManager
{
//...
}
This is to easily distinguish between an interface and classes. In fact, it's a well known standard for defining interfaces.
private readonly ILogger<ClassName> _logger;
private long _rowsAffected;
private IEnumerable<Persons> _people;
This is to easily differentiate between local and global identifiers/variables.
Do declare all member variables and fields at the top of a class, with static fields at the very top:
private static string _externalIdType;
private readonly ILogger<PersonManager> _logger;
private int _age;
This is just a generally accepted practice that prevents the need to hunt for variable declarations.
public class SomeClass
{
public void SomePublicMethodA()
{
}
// rest of your public methods here
private void SomePrivateMethodA()
{
}
private void SomePrivateMethodB()
{
}
}
#region Private Members
private void SomePrivateMethodA()
{
}
private void SomePrivateMethodB()
{
}
#endregion
The preceding code is a code smell which could potentially make your code grow without you realizing it. I admit that I have used this feature many times to collapse the code within a class. However, I realize that hiding code into regions won't give you any value aside from maximizing your visual view when the region is collapsed. If you are working with a team of developers on a project, chances are, other developers will append their code in there until the code get's bigger and bigger over time. As a good practice, it's always recommended to keep your classes small as possible.
If you have tons of private methods within a class, you could split them into a separate class instead.
private readonly CreateQuestionDefinitionRequestDto _requestDto;
It would be too much to name a variable "createQuestionDefinitionRequestDto" when you know that the variable/parameter is a request object. The same thing applies for FTP, UI, IO, etc. It's perfectly fine to use abbreviation for as long as they're generally known, otherwise it would be counter productive not to do so.
public PersonManager person_Manager;
private long rows_Affected;
private DateTime row_updated_date_time;
The reason being is that C# isn't postgres. Seriously, it's to be consistent with the Microsost .NET Framework convention and makes your code more natural to read. It can also avoid "underline stress" or inability to see underline.
public static const string EXTERNALIDTYPE = "ABC";
public static const string ENVIRONMENT_VARIABLE_NAME = "TEST";
They just grab too much attention.
int iCounter;
string strName;
string spCreateUsers;
OrderingService svcOrdering;
Visual Studio code editor already provides helpful tooltips to determine object types. In general, you want to avoid type indicators in the identifier.
The following is an example for defining an enum:
public enum BeerType
{
Lager,
Ale,
Ipa,
Porter,
Pilsner
}
Again, this is to be consistent with the Microsoft .NET framework and avoids type indicators in the identifier.
Try to use record types for immutable objects. Record types is a new feature introduced in C# 9 where it simplfies your code. For example, the following code:
public class Person
{
public string FirstName { get; init; }
public string LastName { get; init; }
public Person(string firstName, string lastName)
{
FirstName = firstName;
LastName = lastName;
}
}
can be written in the following way using record:
public record Person(string FirstName, string LastName);
Using record types will automatically generates the boilerplate code for you and keeping your code concise. Records will be really useful for defining DTOs, Commands or any object that carries immutable data around. For more information about this feature, see: Record Types
before c# 9.0 we used C# compiler directive on local functions
static void Main(string[] args)
{
static void DoAction()
{
// DoAction
Console.WriteLine("DoAction...");
}
#if DEBUG
DoAction();
#endif
}
in c# 9.0 can be written in the following way using Attributes on local functions:
static void Main(string[] args)
{
[Conditional("DEBUG")]
static void DoAction()
{
// Do Action Here
Console.WriteLine("Do Action...");
}
DoAction();
}
Visual Studio shortcut : CRTL+K+D
if (RequestIsValid(request.Id))
{
}
bool RequestIsValid(int id)
{
return id > 0;
}
Visual Studio shortcut : CRTL+R+G
public IEnumerable<User> GetUsers()
{
if (SomeConditions())
{
return null;
}
return Data;
}
đź‘Ť Do use Enumerable.Empty instead:
public IEnumerable<User> GetUsers()
{
if (SomeConditions())
{
return Enumerable.Empty<User>();
}
return Data;
}