A few days ago a friend pointed out to me that I had a wrong idea of composition in UML. She was completely right, so I decided to find out what more I could have been wrong about. Right now, there is one more thing that I have doubts about: I have a circular dependency in my codebase that I would like to present in UML form. But how.
In my case the following is true:
Both A and B have a list of C
C has a reference to both A and B to get information from.
C cannot exist if either A or B stops to exist
Both A and B remain to exist after C is deleted from A and/or B
To model this, I've come up with the following UML (I've ommited multiplicities for now, to not crowd the diagram.)
My question is, is this the right way to model such relations?
Problems
Some facts to keep in mind:
Default multiplicity makes your model invalid. A class may only be composed in one other class. When you don't specify multiplicity, you get [1..1]. That default is sad, but true.
The UML spec doesn't define what open-diamond aggregation means.
Your model has many duplicate properties. There is no need for any of the properties in the attribute compartments, as there are already unnamed properties at the ends of every association.
Corrections
Here is a reworking of your model to make it more correct:
Notice the following:
The exclusive-or constraint between the associations means only one of them can exist at a time.
Unfortunately, the multiplicities allow an instance of C to exist without being composed by A or B. (See the reworked model below.)
The property names at the ends of all associations explicitly name what were unnamed in your model. (I also attempted to indicate purpose in the property names.)
The navigability arrows prevent multiple unwanted properties without resorting to duplicative attributes.
Suggested Design
If I correctly understand what your model means, here is how I would probably reverse the implementation into design:
Notice the following:
Class D is abstract (the class name is in italics), meaning it can have no direct instances.
The generalization set says:
An instance cannot be multiply classified by A and B. (I.e., A and B are {disjoint}.)
An instance of D must be an instance of one of the subclasses. (I.e., A and B are {complete}, which is known as a covering axiom.)
The subclasses inherit the ownedC property from class D.
The composing class can now have a multiplicity of [1..1], which no longer allows an instance of C to exist without being composed by an A or a B.
Leave away the open diamonds and make them normal associations. These are no shared aggregations but simple associations. The composite aggregations are ok.
In general there is not much added value in showing aggregations at all. The semantic added value is very low. In the past this was a good hint to help the garbage collection dealing with unneeded objects. But nowadays almost all target languages have built-in efficient garbage collectors. Only in cases where you want an explicit deletion of the aggregated objects you should use the composite aggregation.
Considerer this:
As far as I know in case 1 a is related to b, and a is aware of b so a is able to send messages to b, but b is not aware of a so b is not able to send messages to a.
In the other hand, in case 2 a is related to b, and both of them are aware of each other, both of them are able to send messages to the other.
But my question is what about that when the relationship is not association but aggregation? Could an aggregation relationship have bidirectional navegability in UML so both instances are aware of each other?
In case a owns b, consequently a is aware of b. (in other words a is able to send messages to b). but what about the other way around, does b is aware of a? or does this could never be the case? or this is just not specified in the diagram and this could be both ways? and, in case b has to be aware of a how would that be expressed(I've never seen an aggregation line with an arrowhead in the other end)?
does all of this applies in the same way for composition?
Another thing slightly related to this I Was wondering, could a non bi-directional association have two roles?
As far as I know this could not be possible because a non bi-directional relationship is a relation when just one of the related instances is aware of the other, and a role means how an instance perceive another one, so we need that both instances are aware of each other to be able to have two roles, is this correct?
Directed associations as per UML 2.5:
A DirectedRelationship represents a relationship between a collection of source model Elements and a collection of target model Elements.
So this does not say much. And in fact you can simply leave it away. There is a concept of non-navigability which explicitly rules out navigation.
Ownership is not shown by an arrow but by a dot near the owning class.
Another cite from Superstructures (p. 200 in chap. 11.5 Associations):
Navigability notation was often used in the past according to an informal convention, whereby non-navigable ends were assumed to be owned by the Association whereas navigable ends were assumed to be owned by the Classifier at the opposite end. This convention is now deprecated. Aggregation type, navigability, and end ownership are separate concepts, each with their own explicit notation. Association ends owned by classes are always navigable, while those owned by associations may be navigable or not.
If you specify a role name that explicitly means you have navigability towards the named class.
I'm reading a book by Eric Evans DDD.
And I found a contradiction.
Chapter books about aggregates:
Choose one ENTITY to be the root of each AGGREGATE, and control all
access to the objects inside the boundary through the root.
Chapter books about repositories:
A subset of persistent objects must be globally accessible through a
search based on object attributes. Such access is needed for the roots
of AGGREGATES that are not convenient to reach by traversal. They are
usually ENTITIES, sometimes VALUE OBJECTS with complex internal
structure, and sometimes enumerated VALUES. Providing access to other
objects muddies important distinctions.
Provide REPOSITORIES only for AGGREGATE roots that actually need
direct access.
It can be concluded that the root of the aggregate can be:
entity
value object
enumerated values
Correctly I understood everything?
Or may be right:
Provide REPOSITORIES only for
aggregate roots
value objects
enumerated values
?
And what is enumerated values(which needs its own repository!)?
Per #Marco's comment above, the root of an aggregate can only be an entity (i.e. something with an ID property). An example of this would be an Order object. No matter how many attributes you change on an Order its quality is determined by its Id property and nothing else.
A value object (often implemented as a struct in many languages) does not have an ID. A common example of this would be a Money value object with a Dollars property and Cents property. Because it has no ID, the concept of querying it by ID does not apply, and thus the concept of a repository does not apply. An aggregate could have a value object as a property, though (e.g. the Total property on an Order aggregate).
An enumerated type is just a list of name/value pairs. It uses the enum keyword in several languages. Again, there's no ID for the enum nor any of its members, so the concept of a repository does not apply. The concept of an enum is useful in DDD because it helps express the domain model better than, say, magic numbers e.g. order.Status = OrderStatus.Submitted vs order.Status = 1.
Entities and value objects are both domain objects. What's the use of knowing the distinction between the two in DDD? Eg does thinking about domain objects as being either an entity or value object foster a cleaner domain model?
Yes, it is very helpful to be able to tell the difference, particularly when you are designing and implementing your types.
One of the main differences is when it comes to dealing with equality, since Entities should have quite different behavior than Value Objects. Knowing whether your object is an Entity or a Value Object tells you how you should implement equality for the type. This is helpful in itself, but it doesn't stop there.
Entities are mutable types (at least by concept). The whole idea behind an Entity is that it represents a Domain concept with a known lifetime progression (i.e. it is created, it undergoes several transformations, it is archived and perhaps eventually deleted). It represents the same particular 'thing' even if months or years pass by, and it changes state along the way.
Value Objects, on the other hand, simply represent values without any inherent identity. Although you don't have to do this, they lend themselves tremendously well to be implemented as immutable types. This is very interesting because any immutable type is by definition thread-safe. As we are moving into the multi-core age, knowing when to implement an object as an immutable type is very valuable.
It also helps a lot in unit testing when the equality semantics are well-known. In both cases, equality is well-defined. I don't know what language you use, but in many languages (C#, Java, VB.NET) equality is determined by reference by default, which in many cases isn't particularly useful.
I have just started reading DDD. I am unable to completely grasp the concept of Entity vs Value objects.. Can someone please explain the problems (maintainability, performance.. etc) a system could face when a Value object is designed as a Entity object? Example would be great...
Reduced to the essential distinction, identity matters for entities, but does not matter for value objects. For example, someone's Name is a value object. A Customer entity might be composed of a customer Name (value object), List<Order> OrderHistory (List of entities), and perhaps a default Address (typically a value object). The Customer Entity would have an ID, and each order would have an ID, but a Name should not; generally, within the object model anyway, the identity of an Address probably does not matter.
Value objects can typically be represented as immutable objects; changing one property of a value object essentially destroys the old object and creates a new one, because you're not as concerned with identity as with content. Properly, the Equals instance method on Name would return "true" as long as the object's properties are identical to the properties of another instance.
However, changing some attribute of an entity like Customer doesn't destroy the customer; a Customer entity is typically mutable. The identity remains the same (at least once the object has been persisted).
You probably create value objects without realizing it; anytime you are representing some aspect of an Entity by creating a fine-grained class, you've got a value object. For example, a class IPAddress, which has some constraints on valid values but is composed of simpler datatypes, would be a value object. An EmailAddress could be a string, or it could be a value object with its own set of behaviors.
It's quite possible that even items that have an identity in your database don't have an identity in your object model. But the simplest case is a composite of some attributes that make sense together. You probably don't want to have Customer.FirstName, Customer.LastName, Customer.MiddleInitial and Customer.Title when you can compose those together as Customer.Name; they'll probably be multiple fields in your database by the time you think about persistence, but your object model doesn't care.
Any object that is collectively defined by all of it attributes is a value object. If any of the attributes change you have a new instance of a value object. This is why value objects are defined as immutable.
If the object is not fully defined by all of its attributes then there are a subset of attributes that make up the identity of the object. The remaining attributes can change without redefining the object. This kind of object cannot be defined at immutable.
A simpler way of making the distinction is to think of value objects as static data that will never change and entities as data that evolves in your application.
Value Types :
Value types do not exist on his own, depends on Entity types.
Value Type object belongs to an Entity Type Object.
The lifespan of a value type instance is bounded by the lifespan of the owning entity instance.
Three Value types: Basic(primitive datatypes), Composite(Address) and Collection(Map, List, Arrays)
Entities:
Entity types can exist on his own (Identity)
An entity has its own life-cycle. It may exist independently of any other entity.
For example: Person, Organisation, College, Mobile, Home etc.. every object has its own identity
I don't know if the following is correct, but I would say that in the case of an Address object, we want to use it as a Value Object instead of an Entity because changes to the entity would be reflected on all linked objects (a Person for instance).
Take this case: You are living in your house with some other people. If we would use Entity for Address, I would argue that there would be one unique Address that all Person objects link to. If one person moves out, you want to update his address. If you would update the properties of the Address Entity, all people would have a different address. In the case of a Value Object, we would not be able to edit the Address (since it is immutable) and we would be forced to provide a new Address for that Person.
Does this sound right? I must say that I was/am also still confused about this difference, after reading the DDD book.
Going one step further, how would this be modelled in the database? Would you have all properties of the Address object as columns in the Person table or would you create a separate Address table that would also have a unique identifier? In the latter case, the people living in the same house would each have a different instance of an Address object, but those objects would be the same except for their ID property.
address can be entity or value object that depends on the busiess process. address object can be entity in courier service application but address can be value object in some other application. in courier application identity matters for address object
3 distinction between Entities and Value Objects
Identifier vs structural equality:
Entities have identifier,entities are the same if they have the same
identifier.
Value Objects on beyond the hand have structural equality, we consider two
value objects equal when all the fields are the same. Value objects cannot
have identifier.
Mutability vs immutability:
Value Objects are immutable data structures whereas entities change during
their life time.
Lifespan: Value Objects Should belong to Entities
In a very simple sentence I can say, we have three types of equality:
Identifier equality: a class has id filed and two objects are compared with their id field value.
Reference equality: if a reference to two objects has a same address in memory.
Structural equality: two objects are equal if all members of them are matched.
Identifier equality refers only to Entity and structural equality refers to Value Object only. In fact Value Objects do not have id and we can use them interchangeably. also value objects must be immutable and entities can be mutable and value objects will not have nay table in database.
I asked about this in another thread and I think I'm still confused. I may be confusing performance considerations with data modelling. In our Cataloging application, a Customer doesn't change until it needs to. That sounds dumb - but the 'reads' of customer data far outnumber the 'writes' and since many many web requests are all hitting on the 'active set' of objects, I don't want to keep loading Customers time and again. So I was headed down an immutable road for the Customer object - load it, cache it, and serve up the same one to the 99% of (multi-threaded) requests that want to see the Customer. Then, when a customer changes something, get an 'editor' to make a new Customer and invalidate the old one.
My concern is if many threads see the same customer object and it is mutable, then when one thread starts to change it mayhem ensues in the others.
My problems now are, 1) is this reasonable, and 2) how best to do this without duplicating a lot of code about the properties.
Consider the following examples from Wikipedia, in order to better understand the difference between Value Objects and Entities:
Value Object: When people exchange dollar bills, they generally do not
distinguish between each unique bill; they only are concerned about the face
value of the dollar bill. In this context, dollar bills are Value Objects. However,
the Federal Reserve may be concerned about each unique bill; in this context each
bill would be an entity.
Entity: Most airlines distinguish each seat uniquely on every flight. Each seat is
an entity in this context. However, Southwest Airlines, EasyJet and Ryanair do
not distinguish between every seat; all seats are the same. In this context, a seat is
actually a Value Object.