Tuesday, November 17, 2015

Anemic data models (ADM) VS Rich Data Models (RDM) in C#

What are anemic data models?
What are Rich Data models ?
Anemic data model VS Technical concerns
3 big Complaints about Anemic model
Anemic model in disguise of IOC and DI
Conclusion
Further references and reading

What are anemic data models?

Anemic in English means plain so Anemic Models means plain Models.

Plain models are classes without business logic / behaviors. For example below is a simple “Customer” model which just has the data (properties) but it does not have business logic (behavior).

class Customer
    {
        private string _CustomerName;

        public string CustomerName
        {
            get { return _CustomerName; }
            set { _CustomerName = value; }
        }
        private string _CustomerCode;

        public string CustomerCode
        {
            get { return _CustomerCode; }
            set { _CustomerCode = value; }
        }

    }

Now in real world plain models are of no use without business logic unless you are creating DTO ( data transfer objects) which are meant for different purpose. Some developers feel comfortable in pushing these actions and methods in a separate classes which are named with different vocabularies like service, façade , manager , utility etc. These plain models are then passed to these classes who run the behavior over them.

For example below is a simple service class which validates the “Customer” object and decides if the object valid or not. Putting in simple words Anemic model has behavior in separate service / utility / manager classes.

class CustomerValidationService
    {
        public bool Validate(Customer obj)
        {
            if (obj.CustomerCode.Length == 0)
            {
  return false;
            }
            if (obj.CustomerName.Length == 0)
            {
                return false;
            }
                return true;
        }
    }

What are Rich Data models ?

Rich data models have both data and behavior in the same class. Below is the RDM for the previously defined ADM. You can see the data ( CustomerCode and CustomerName) and behavior (Validate) both are part of the same model.

class Customer
    {
        private string _CustomerName;

        public string CustomerName
        {
            get { return _CustomerName; }
            set { _CustomerName = value; }
        }
        private string _CustomerCode;

        public string CustomerCode
        {
            get { return _CustomerCode; }
            set { _CustomerCode = value; }
        }

        public bool Validate()
        {
                if (CustomerCode.Length == 0)
                {
                    return false;
                }
                if (CustomerName.Length == 0)
                {
                    return false;
                }
                return true;
          
            
        }
        
    }

Anemic data model VS Technical concerns

Lot of developers confuse Technical cross cutting concerns with Anemic model. In Anemic model the behaviors in the service class is the behavior of the real world model.

Lot of developers create utility or service classes which decouple technical behavior from the model which is not anemic model. For example someone can create a “Dal” class as shown below and pass the “Customer” object to be inserted in to the database. Lot of people term this kind of approach as “Repository” pattern.

class Dal
{
        public void InsertDb(Customer obj)
        {
            // Code of ADO.NET goes here
        }
}

Now this kind of model is not anemic model. This is separating the technical behavior from the business behavior which perfectly makes sense. In the previous example we had “validate” method which actually belongs to the “Customer” model. So Anemic model are those models which have business behavior separated in a different class and not technical behavior. So the above “Dal” class is not an Anemic model.

Customer cust = new Customer();
cust.CustomerName = "Shiv";
Dal dal = new Dal();
dal.InsertDb(cust);

3 big Complaints about Anemic model

Lot of developers unknowingly follow Anemic model because of simplicity, but OOP community has the below concerns for ADM:-

  • Anemic model disrespects OOP concepts. Objects in OOP have data and behavior. Object should model real time entity. By separating the behavior we are not following OOP. With behaviors residing in a separate class it would be difficult to inherit , apply polymorphism , abstraction and so on.
     
  • Anemic models can have inconsistent states at any time. For example see the below code we have first created a valid customer object and the validation service logic ran over it and set the “IsValid” to true. So till step 1 everything is fine “IsValid” is true and in synch with the customer object data. Now somewhere down the line you can say in step 2 customer name property is set to nothing. In this step the customer object is not valid in reality but if you try to display in Step 3 it shows valid. In order words the data and the IsValid flag is not in synch
Customer obj = new Customer();
obj.CustomerName = "Shiv";
obj.CustomerCode = "1001";

CustomerValidationService validationService = new CustomerValidationService();
validationService.Validate(obj);
Console.WriteLine(obj.IsValid); // Step 1 :- till here Valid

obj.CustomerName = ""; // Step 2 :-  we made the model invalid
Console.WriteLine(obj.IsValid); // Step 3 :- but it says Still Valid

Anemic model in disguise of IOC and DI

With all respect to the views of OOP developers Anemic models exists very beautifully and amicably in the disguise of DI IOC architecture today. I personally find it very cruel to say it’s an Anti-pattern.

Consider the below scenario where we have a customer class who has different kind of discount calculations. So rather than putting the discount calculation logic inside the customer class you can inject the object from the constructor as shown in the below. In case you are new to dependency injection please read this DI IOC in C#.

class Customer
    {
        private IDiscount _Discount = null;
        public Customer(IDiscount _Dis)
        {
            _Discount = _Dis;
        }
        private string _CustomerName;

        public string CustomerName
        {
            get { return _CustomerName; }
            set { _CustomerName = value; }
        }

        public string Region { get; set; }
        public double Amount { get; set; }
        public double CalculateDiscount()
        {
            return _Discount.Calculate(this);
        }
        
    }

And your discount calculation logic is in a different class all together which takes the customer object for calculating discount.

public interface IDiscount
{
        double Calculate(Customer Cust);
}

Below are two kinds of discount calculations one is a week end discount and the other is a special discount. The week end discount does calculation as per region and week day while the special discount is a 30% flat.

Now the below classes also follow anemic model because the actions are in a different class but still its perfectly decoupled, valid and OO.

public class WeekEndDiscount : IDiscount
{
        public double Calculate(Customer Cust)
        {
            if (Cust.Region == "Asia")
            {
                if (DateTime.Today.DayOfWeek == DayOfWeek.Sunday)
                {
                    return (Cust.Amount * 0.20);
                }
            }
            return (Cust.Amount * 0.10);
        }
    }
    public class SpecialDiscount : IDiscount
    {

        public double Calculate(Customer Cust)
        {
            return (Cust.Amount * 0.30);
        }
    }

Conclusion

  • Anemic data model is a simple plain model class with only data and behavior resides in a separate class.
  • Rich data model have both data and behavior.
  • Anemic model is criticized for not been OO , extensible , inconsistent and follows procedural programming approach.
  • Anemic model should be confused with cross cutting concern technical classes like Logging , Repository etc.
  • DI IOC indirectly forces anemic model. So Anemic model in today world is very much present in the disguise of DI IOC.

Further references and reading

I have been writing a lot on design patterns, architecture, domain driven development and so on. So below are the link feel free to read and do give feedback.

Thursday, November 12, 2015

Aggregate root pattern in C#

Aggregate root are cluster / group of objects that is treated as single unit of data.

I am sure lots of developers are already using this pattern unknowingly, via this short note I would like to inform you formally what you are doing.

Let us try to understand the above definition with an example. Consider the below “Customer” class which has the capability to add multiple “Address” objects to it. In order to achieve the same we have exposed an address collection from the customer class to represent the 1 to many relationships.

The above class structure works perfectly well. You can create object of customer and add multiple addresses object to it.

Customer cust = new Customer();
cust.CustomerName = "Shiv koirala";
cust.DateofBirth = Convert.ToDateTime("12/03/1977");

Address Address1 = new Address();
Address1.Address1 = "India";
Address1.Type = "Home";
cust.Addresses.Add(Address1);

Address Address2 = new Address();
Address2.Address1 = "Nepal";
Address2.Type = "Home";
cust.Addresses.Add(Address2);

Now let's say we want to implement the following validations:-

“Customer can only have one address of Home type”.

At this moment the address collection is a NAKED LIST COLLECTION which is exposed directly to the client. In other words there are no validations and restrictions on the “Add” method of the list. So you can add whatever and how much ever address objects as you wish.

cust.Addresses.Add(Address2);

So how to address this problem in a clean and logical way. If you think logically “Customer” is composed of Addressescollection, so Customer is like a main root.So rather than allowing DIRECT NAKED ACCESS to Addresses list how about accessing the address list from the customer class.In other words centralizing access to address objects from the customer class.

So below are three steps which I have implemented to put a centralize address validation.

Step 1 :- I have made the address list private. So no direct access to the collection is possible.

Step 2 :- Created a “Add” method in the “Customer” class for adding the “Address” object. In this add method we have put the validation that only one “Home” type address can be added.

Step 3 :- Clients who want to enumerate through the address collection for them we have exposed “IEnumerable” interface.

If you analyze the above solution closely the customer is now the root and the address object is manipulated and retrieved via the customer class.

Why did we do this? , because “Customer” and “Address” object is one logical data unit. To maintain integrity of address validation we need to go via the “Customer” class. In the same way loading of data ,updation , deletion etc should all happen via the “Customer” class so that we have proper data integrity maintained.

So when we say load customer from database all the respective address objects should also get loaded.

So when a group of objects which form one logical unit should have centralized root via which the manipulation of the contained object should happen. This kind of arrangement is terms as “Aggregate Root”.

The best way to Learn Design pattern is by doing a project. So if interested you can start from the below youtube video which demonstrates Design pattern by doing a project.

Friday, November 6, 2015

Data Transfer object design pattern in C#

Introduction and Definition

Scenario 1 :- Improve performance

Scenario 2:- Flatten object hierarchy

Scenario 3:- Exclude properties

Difference between DTO and Business objects

Introduction and Definition

DTO(Data Transfer objects) is a data container for moving data between layers. They are also termed as transfer objects. DTO is only used to pass data and does not contain any business logic. They only have simple setters and getters.

For example below is an Entity class or a business class. You can see that it has business logic in the setters.

class CustomerBO
{
        private string _CustomerName;
        public string CustomerName
        {
            get { return _CustomerName; }
            set 
            {
                if (value.Length == 0)
                {
                    throw new Exception("Customer name is required");
                }
                _CustomerName = value; 
            }
        }

}

A data transfer object of the above Customer entity class would look something as shown below. It will only have setters and getters that means only data and no business logic.

class CustomerDTO
{
        public string CustomerName { get; set; }
}

So in this article let us try to understand in what kind of scenarios DTO’s are used.

Scenario 1 :- Improve performance

Consider the below scenario we have a customer business class and a product business class.

class CustomerBO
{
        private string _CustomerName;
        public string CustomerName
        {
            get { return _CustomerName; }
            set 
            {
                if (value.Length == 0)
                {
                    throw new Exception("Customer name is required");
                }
                _CustomerName = value; 
            }
        }
}
public class ProductsBO
    {
        private string _ProductName;

        public string ProductName
        {
            get { return _ProductName; }
            set 
            {
                if (value.Length == 0)
                {
                    throw new Exception("Product Name is required");
                }
                _ProductName = value; 
            }
        }

        public double ProductCost { get; set; }
    }

Assume there is a remote client which is making calls to get Customer data and the Products they bought. Now the client has to make two calls one to get Customer data and the other to get Products data. Now that quiet inefficient. It would be great if we can get data in one call.

DataAccessLayer dal = new DataAccessLayer();
//Call 1:-  get Customer data
CustomerBO cust = dal.getCustomer(1001);

//Call 2:-  get Products for the customer
ProductsBO prod = dal.getProduct(100);

So to achieve the same we can create a unified simple class which has properties from both the classes.

class CustomerProductDTO
{
  // Customer properties
        public string CustomerName { get; set; }
   // Product properties
        public string ProductName { get; set; }
        public double ProductCost { get; set; }
}

You can now get both Customer and Product data in one go.

//Only one call
CustomerProductDTO cust = dal.getCustomer(1001);

Proxy objects in WCF and Web services are good candidates for data transfer objects to improve performance.

Scenario 2:- Flatten object hierarchy

The second scenario where I end up using DTO objects isfor flattening the object hierarchy for easy data transfer.

For example you have a customer business class which has one to many relationships with an address class. But now let us say we want to write this whole data to a CSV file or transfer it across layers. In those scenarios a flattened denormalized structure makes things easier.

So you have a customer class which has one to many relationship with address objects. It also has business rules in it.

class CustomerBO
{
 // Customer properties removed for reading clarity
public List Addresses { get; set; }
}
class AddressBO
    {
private string _Address1;

public string Address1
        {
get { return _Address1; }
set
            {
if (value.Length == 0)
                {
throw new Exception("Address is required");
                }
                _Address1 = value; 
            }
        }

    }

To create a DTO of the above customer and address class we can just merge both the properties of the classes in to one class and exclude the business logic.

class CustomerDTO
{
public string CustomerName { get; set; }
public string ProductName { get; set; }
public double ProductCost { get; set; }
}

Scenario 3:- Exclude properties

Because DTO is completely detached from the main business object you have the liberty to remove certain fields which you do not want to transfer to the other layers. For example below is a simple customer business object class which has two properties called as “IsAdmin” and “CustomerName”.

When you pass this business object to the UI layer you do not wish to transfer the “IsAdmin” value to the UI.

class CustomerBO
{
 private bool _IsAdmin;
        public string IsAdmin
        {
            get { return _IsAdmin; }
        }
        private string _CustomerName;
        public string CustomerName
        {
            get { return _CustomerName; }
            set 
            {
                if (value.Length == 0)
                {
                    throw new Exception("Customer name is required");
                }
                _CustomerName = value; 
            }
        }
}

So you can create a DTO which just has the “CustomerName” property.

class CustomerDTO
{
        public string CustomerName { get; set; }
}

Difference between DTO and Business objects

  Business objects DTO
Data Yes Yes
Business logic Yes No
Structure Normalized Normalized or Denormalized

If we can mathematically summarize:-

Business object = Data + Logic
                 DTO = Data

In case you want to learn design pattern I would suggest to learn design pattern with a project. Do not learn each design pattern individually. Because when design patterns are used in project they overlap with each other and such kind of experience can only be felt by doing an actual project and implementing design patterns on demand and naturally.

Below is a youtube video which demonstrates 15 important design pattern in a C# project.

Sunday, November 1, 2015

Value Object design pattern in C#

Introduction of Value object design pattern

Definition: - “Value object is an object whose equality is based on the value rather than identity. “

Let us understand the above statement with more clarity. When you create two objects and even if their values are same they represent different entities. For example in the below code we have created two person objects with the same name “Shiv”.

Person PersonfromIndia = new Person();
 PersonfromIndia.Name = "Shiv";
 PersonfromIndia.Age = 20;
 
 Person PersonfromNepal = new Person();
 PersonfromNepal.Name = "Shiv";
 PersonfromNepal.Age = 20;
 
But the first person stays in “India” and the other stays in “Nepal”. So in other words “PersonFromIndia” object is different from “PersonFromNepal” object even if the person’s name and age is same. In other words they have DIFFERENT IDENTITIES.

If you try to compare the above C# object it will return false and this is completely in line with our expectations. You can use the “Equal” method or you can use “==”.



if (PersonfromIndia.Equals(PersonfromNepal))
 {
                
 }
 
But now consider the below scenario of a money example. We are creating two money objects of 1 rupee value but one is using a paper material and the other is made of steel material.



But in this case “OneRupeeCoin” value is equal to “OneRupeeNote” value. So even if the objects are of different instances they are equal by money value. So if the money and currency type matches both the objects are equal.

Money OneRupeeCoin = new Money();
 OneRupeeCoin.Value = 1;
 OneRupeeCoin.CurrencyType = "INR";
 OneRupeeCoin.CurrencyType = "INR";
 OneRupeeCoin.SerialNo = "A3123JJK332";
 
 Money OneRupeeNote = new Money();
 OneRupeeNote.Value = 1;
 OneRupeeCoin.CurrencyType = "INR";
 OneRupeeCoin.CurrencyType = "INR";
 OneRupeeNote.Material = "Paper";
 OneRupeeCoin.SerialNo = "Z2232V4455";

In other words value objects when compared are same when the values of the properties are same.

Implementing Value object pattern in C# is a two step process , so in the further article let us run through those steps.

Step 1 :- Making Value object pattern work logically in C#

Now for the above “Money” value object to function properly the below comparison should work both for “Equals” and “==”. But technically C# does object reference comparison and not value.



if (OneRupeeCoin==OneRupeeNote)
 {
 Console.WriteLine("They are equal");
 }
 if (OneRupeeCoin.Equals(OneRupeeNote))
 {
 Console.WriteLine("They are equal");
 }

So to achieve the same we need to override the “equals” methods and overload “==” operator as shown in the code below. You can see now the equality is compared on the base of “Value” and “CurrencyType”. If the “CurrencyType” and “Value” is same that means the objects are same.


class Money
     {
         public override bool Equals(object obj)
         {
             var item = obj as Money;
             if (item.Value == Value)
             {
                 return true;
             }
             return false;
         }
         public static bool operator !=(Money money1, Money money2)
         {
             if ((money1.Value == money2.Value) &&
                 (money1.CurrencyType == money1.CurrencyType))
             {
                 return true;
             }
             return false;
         }
         public static bool operator ==(Money money1, Money money2)
         {
             if ((money1.Value == money2.Value)&&
                 (money1.CurrencyType == money1.CurrencyType))
             {
                 return true;
             }
             return false;
         }
     }

Once the above methods are incorporated the equality will work on the values and not the reference. This is in synch with the Value object pattern behavior we discussed in the introduction.

Step 2 :- Value objects should be immutable

Now let us say we are using the above money object inside a product class. So you can see we have created two product objects one for shoes and one for chocolates.

But BY MISTAKE we have assigned the same “Money” object to both the products. In other words the money object is now shared between both the products.



Product shoes = new Product();
 shoes.ProductName = "WoodLand";
 
 // Step 1:- Money object set to 100 INR
 Money Cost = new Money();
 Cost.CurrencyType = "INR";
 Cost.Value = 100;
 shoes.Cost = Cost;
 
 // Step 2 :- the same money  object is modified to 1 value
 // this affected the shoes cost as well
 Product Chocolates = new Product();
 Chocolates.ProductName = "Cadbury";
 Chocolates.Cost = Cost;
 Chocolates.Cost.Value = 1;
 Chocolates.Cost.CurrencyType = "USD";
 
So in Step 1 the money object was set to 100 INR and in step 2 it was modified to 1 USD. Step 2 affected the cost of Shoes product as well. This can cause lot of defects and unusual behavior.

A Value “100” cannot be replaced by anything, “100” is “100”. In other words value objects should be made IMMUTABLE to avoid confusion. In case you are new to the concept of immutability I would suggest reading this article on C# Immutable objects.

Immutable means once the object is filled with data, that data cannot be changed.

To make the money class immutable is a 2 step process:-
  • Make the property set as private.
  • Make provisions to set data via constructor.
Below is the code with comments of both the steps.


class Money
 {
         // Step 1 :- Make the sets private
         public readonly double Value { get; private set; }
         public readonly string CurrencyType { get; private set; }
         public readonly string Material { get; private set; }
         
 // Step 2 :- Pass data via constructor
         public Money(double _Value,
                      string _CurrencyType,
                     string _Material)
         {
             Value = _Value;
             CurrencyType = _CurrencyType;
             Material = _Material;
         }
 // Code of Equals and == operator removed for simplification
 }
 
Below is the code which shows how the product object will be assigned with money immutable object. This would further avoid changes to the money object and any confusion around the same. Now the money value object of shoes is different from chocolates.


Product shoes = new Product();
 shoes.ProductName = "WoodLand";
 shoes.Cost = new Money(100, "INR", "Paper");
 
 Product Chocolates = new Product();
 Chocolates.ProductName = "Cadbury";
 Chocolates.Cost = new Money(1, "USD", "Coin");
 


Conclusion about Value object pattern


  • Value objects equality is based on value rather than identity.
  • Value objects should me IMMUTABLE to avoid confusion.
  • In C# to ensure proper behavior of value object we need to override “Equals” method and “==” operator.
The best way to learn design pattern is by doing a project. Below is a simple customer project in which I have implemented 8 design patterns (Factory Pattern,Lazy Loading,RIP Pattern,Stratergy Pattern,Decorator pattern,Repository,Unit of Work and Template pattern).

Below is the first video of Learn Design pattern in 8 hours.