I have an MVC2 n-tier application (DAL, Domain, Service, MVC web) using a DDD approach (Domain Driven Design), having a Domain Model with repositories. My service layer uses a Request/Response pattern, in which the Request and Response objects contain DTO's (Data Transfer Objects) to marshal data from one layer to the next, and the mapping is done via help from AutoMapper. My question is this: what shape should a DTO typically take? Can it have nested/complex DTO's as well or should it strictly be a flat projection? Or possibly a mixture of both? Also, what are the main reasons for having a flat DTO vs a more complex/nested DTO?
For instance, suppose I had a domain such as the following:
public class Employee
{
public string FirstName { get; set; }
public string LastName { get; set; }
public Company Company { get; set; }
}
public class Company
{
public string Name { get; set; }
public string Address { get; set; }
public string City { get; set; }
public string State { get; set; }
}
There are three different ways I've thought of modeling the Response object.
Option 1 - the DRYest option:
public class GetEmployeeResponse
{
public class EmployeeDTO { get; set; } // contains a CompanyDTO property
}
From the research I've done, it would be inappropriate for a DTO to take a similar shape as the domain object(s) as demonstrated above.
Option 2 - a flattened projection of the domain (anti-DRY):
public class GetEmployeeResponse
{
public string FirstName { get; set; }
public string LastName { get; set; }
public string CompanyName { get; set; }
public string CompanyAddress { get; set; }
public string CompanyCity { get; set; }
public string CompanyState { get; set; }
}
This is more simple, like a DTO apparently should be, but ultimately makes for more DTOs.
Option 3 - a mixture of both:
public class GetEmployeeResponse
{
public EmployeeDTO Employee { get; set; }
public CompanyDTO Company { get; set; }
}
This allows for the code to be a little bit more dry, reusable and manageable, and doesn't expose my domain structure to the end user. The other main benefit is that other responses, like GetCompanyResponse could simply return CompanyDTO, without having to make a copy of all those properties, similar to option 2. What do you think? Which option of these (if any) have you taken and/or have worked for you? If these Request/Responses later get exposed as WCF service methods, does your answer change?
My personal preference would be to try and keep it flat as possible with only the required data being transfered. having said that I have used deeply nested DTO in the past because it made sense at the time and fitted the requirements. so I guess it comes down to "it depends". At the end of the day go with what makes sense for the application at hand. No point trying to shoe horn data into a DTO convention that doesn't fit what you are tying to achieve.
Related
I am making ASP.NET core MVC project. In one Solution I have some assemblies - .Data, .Services, .Web. In the .Data there are the Models (for the database), DbContext, Migration etc. For the models I use DataAnnotations, but when there are some "magic" numbers like 100 (for the maximum length of the "Name" property) I prefer to use constants.
For example [MaxLength(CourseConstants.NameMaxLength)]. These constants I can use them in the ".Web" assemblie. For example when someone is making a course and I can restrict the maximum length of the name using the same constant like in the "Course" model.
So my question is - where to put the constants ? One way is in the separate class (example - "CourseConstants"). But I think that for better cohesion I should put the constants that are used for the "Course" model inside the "Course" class. And if I need them in ".Services" or in ".Web" I should call them from the Course class - example Course.NameMaxLength.
P.S. sorry if I dont follow some of the rules in this site and community, its my first question here.
CASE 1:
public class Course
{
public int Id { get; set; }
[Required]
[MaxLength(CourseConstants.NameMaxLength)]
public string Name { get; set; }
[Required]
[MaxLength(CourseConstants.DescriptionMaxlenght)]
public string Description { get; set; }
public DateTime StartDate { get; set; }
public DateTime EndDate { get; set; }
[Required]
public User Trainer { get; set; }
public string TrainerId { get; set; }
}
public class CourseConstants
{
public const int NameMaxLength = 100;
public const int DescriptionMaxlenght = 200;
}
CASE 2:
public class Course
{
public const string NameMaxLength = 100;
public const string DescriptionMaxlenght= 200;
public int Id { get; set; }
[Required]
[MaxLength(NameMaxLength)]
public string Name { get; set; }
[Required]
[MaxLength(DescriptionMaxlenght)]
public string Description { get; set; }
public DateTime StartDate { get; set; }
public DateTime EndDate { get; set; }
[Required]
public User Trainer { get; set; }
public string TrainerId { get; set; }
}
A common reason to separate models in a different project is that you might want to reuse them for other applications. For example you have an API and it returns some models in JSON format and you would want to use the exact same model in an application that consumes that API, since it's the same. Even better, when someone changes the class, it gets updated in your app as well.
Your question, though, relates to data models used inside the data access layer. These models should only be used in your business classes and, of course, in the DAL. They should not be reusable outside the scope of the DAL and projects that access data. If you want to make them consumable from the outside, you would map them to API models that would be different classes than EF classes.
That said, constants should be part of the DAL project, together with the EF data classes, since they should not be reused anywhere else. The only reason why you would put them somewhere else would be as an assembly to share between projects with constants like PersonNameLength or EmailLength or something like that.
I'm practicing to implement a project using CQRS and DDD to the best I can and have come up with a question once building the commands.
Scenario: The user sends a command to create a bookshelf in the system which itself could also contain a collection of books.
The command looks like this:
public class CreateNewBookShelfCommand : ICommand
{
public long CommandInitiatorId { get; set; }
public string Name { get; set; }
public string Description { get; set; }
public ICollection<Book> Books { get; set; }
}
All of the Commands reside in the Contracts project that itself is inside the Application solution folder.
The Model project residing in the Domain solution folder hosts the BookShelf entity class which includes a collection of Books.
Questions 1: The command above has this Books property which is of type Book. My question is: should the Commands project directly reference the Model project so that the Book data type would be resolved in the CreateNewBookShelfCommand? Myself, I don't think that the Contracts are allowed to reference any project other than the CommandHandlers or anything else which might be Cross Cutting.
Question 2: So, is this a good practice to replicate part of the Book entity class here at the Contracts project and utilize it in the CreateNewBookShelfCommand?
And this is the Model project I have for Book and BookShelf:
public class BookShelf : BaseEntity
{
public string Name { get; set; }
public string Description { get; set; }
public BookShelfAccess Access { get; set; }
public virtual BookShelfOwner Owner { get; set; }
public long OwnerId { get; set; }
public ICollection<Book> Books { get; set; }
}
public class Book : BaseEntity
{
public string Name { get; set; }
public DateTime? PublishedAt { get; set; }
public int NumberOfPublishes { get; set; }
public virtual BookShelf Shelf { get; set; }
public long ShelfId { get; set; }
}
I hope my explanation has been enough if there is any other information I should add here let me know.
Typically, you would not use the entities in your domain model as elements of your command.
Commands are fundamentally messages, and as such they really ought to be immutable. You want to know that what is received is the same thing that was sent. In DDD terms, you might think of the messages as being value objects.
It doen't make sense to use an entity from your domain model as part of the in memory representation of your command, because you should never be invoking any of the entity methods that change its state.
Commands are much closer in nature to data transfer objects than they are to domain model entities.
Normally, all of the entity objects in your system should live behind an aggregate root interface, and the arguments that you pass to that interface are values. If the domain model needs an entity, it can create one from the values provided.
I'm quite familiar with RIA Services, but not with WCF Data Services, and I'm wondering how to do something in the latter which I know how to do in the former.
I know examples for each data access methods on how to query database entities when the data service is backed by entity framework (or something similar).
But often enough I want to offer a queryable service method that isn't based on any database entity directly. It could be a combination from several database entities, an inherent grouping, or simply a database entity together with some additional, computed data.
Let's say I have the following classes:
// part of the model
class SomeEntity
{
public Guid Id { get; set; }
public String PropertyA { get; set; }
public String PropertyB { get; set; }
public Guid SubsidiaryKey { get; set; }
public virtual SomeSubsidiaryEntity Subsidiary { get; set; }
public virtual ICollection<SomeRelatedEntity> RelatedEntities { get; set; }
}
// part of the model
class SomeSubsidiaryEntity
{
public Guid Id { get; set; }
public String PropertyS { get; set; }
}
// not part of the model, exists only for the service layer
class SomeEntityProjection
{
public Guid Id { get; set; }
public String PropertyA { get; set; }
public String PropertyB { get; set; }
public String PropertyS { get; set; }
public Int32 RelatedEntitiesCount { get; set; }
}
The first two are part of the database model the service is using, the last is a projection entity.
The projection entity isn't part of the database - it only exists in the service layer.
With RIA Services, I can now write a query function like this:
public IQueryable<SomeEntityProjection> GetSomeEntitiesWithSomeFluff()
{
return
from e in this.DbContext.SomeEntities
select new SomeEntityProjection()
{
PropertyA = e.PropertyA,
PropertyB = e.PropertyB,
PropertyS = e.Subsidiary.PropertyS,
RelatedEntitiesCount = e.RelatedEntities.Count()
}
}
The data service method then supports client-side filtering and sorting for all four properties. All such client-side sorting and filtering will be properly delegated to the database and done there.
I couldn't find any examples of this for WCF Data Services, so my question is: Is this possible with WCF Data Services also and if so, how to do it?
You would use Data Transfer Objects (DTO) across the wire, which are similar, but not necessarily identical to your entity objects. That way they can vary independently.
I have a model and a partial model which contains only the properties that I need to expose in JSON.
But the properties between the model and his partial model are redundant.
How can I avoid that or improve my approach?
namespace Dashboard.Models.UserModels
{
public class UserModel
{
public int id { get; set; }
public string dbName { get; set; }
public string firstname { get; set; }
public string lastname { get; set; }
public int idExternal { get; set; }
public int idInstance { get; set; }
public string login { get; set; }
public string password { get; set; }
public DateTime? dtContractStart { get; set; }
public DateTime? dtContractEnd { get; set; }
public string emailPro { get; set; }
public string emailPerso { get; set; }
public LuccaUserModel()
{
idInstance = -1;
}
// partial model for json result
// not sure is the best way or have to be here
public class PartialUserModel
{
public int id { get; set; }
public string firstname { get; set; }
public string lastname { get; set; }
public string emailPro { get; set; }
public string emailPerso { get; set; }
public DateTime? dtContractStart { get; set; }
public DateTime? dtContractEnd { get; set; }
public string url { get; set; }
}
// UserModel Methods
}
}
You can rename PartialUserModel UserModelBase class (or leave it as is... it just makes better logical sense to do so) and make UserModel to inherit from it:
public class UserModel : UserModelBase
{
...
}
Of course you'll need to remove all duplicate properties from UserModel in this case.
It's a thin line between doing a proper design and building an overkill design. Answer depends on many inputs, among which I chose to have project and model breadth most important.
In hope to have my answer clearer, I have to say I use different terminology. Data which is adopted for use in UI is usually called ViewModel. In your case, you would build UserViewModel which contains necessary subset of information.
If I'm working on a one-off project, I'll reuse model as a ViewModel. I'll do this by having helper method which removes sensitive information, loads up or cuts off data which is lazy loaded from database and does other preparation on data. All this is done with same model class.
If it's not a short term project, I look to create separate ViewModel classes which I map from model data. Then, if I'm working with mostly flat data I use AutoMapper tool to have data automatically copied, instead of writing my own mappers.
As another answer here states, you write a basic class with data you need in UI and extend it with other model data, however this is not a good approach for several reasons.
If violates separation of concerns. Project dealing with model and persistance should not know about your ViewModel
You may need to flatten data from related objects into ViewModel objects. In that case, your model objects would have fields which should not be there, or would be redundant.
You may need calculated fields and helper methods in ViewModel which would again end up in model, confusing everyone that is not updated about design.
You could want to adopt several unrelated model classes to same ViewModel class
To try and put it shortly, either reuse model class or create ViewModels. There is unfortunately no clever solution. If you find one, please post a comment as I'd like to hear about it :)
1) I want to know what is the recommended way to create & return a DTO for an object which has 10 attributes and I only want to return 2 with my DTO object.
2) Should DTO's have their own namespace ? If yes, how do we organize them ? Each DTO inside a single class file or all DTO's inside a single class ?
Please provide me some sample code.
DTOs are dumb objects composed of public getters/setters. I generally put them in a separate namespace called SomeProject.Dto.
public class CustomerDto {
public int Id { get; set; }
public string Name { get; set; }
public LocationDto HomeAddress { get; set; }
}
I generally try to keep the property names the same between the DTO and corresponding domain class, possibly with some flattening. For example, my Customer might have an Address object, but my DTO might have that flattened to:
public class CustomerDto {
public int Id { get; set; }
public string Name { get; set; }
public string HomeStreet { get; set; }
public string HomeCity { get; set; }
public string HomeProvince { get; set; }
public string HomeCountry { get; set; }
public string HomePostalCode { get; set; }
}
You can dramatically reduce the amount of repetitive mapping code of translating domain objects into DTOs by using Jimmy Bogard's AutoMapper.
http://automapper.codeplex.com/
Your question is very open ended. The answers are dependent on the scale of your application.
In general I create my DTO's or ViewModels in their own assembly. To get my DTO's I have some service layer take care of creating them based on my request.
If you want concrete examples take a look at some of the Asp.NET MVC examples at asp.net. While you may not be using MVC you can at least see how the ViewModels are created.