I've been reading a lot of blogs which advocate the fat models and skinny controllers approach, esp. the Rails camp. As a result the routers is basically just figuring out what method to call on what controller and all the controller method does is call the corresponding method on the model and then bring up the view. So I've two concerns here which I don't understand:
The controller and router are really not doing much different tasks other than just calling a method on the God-like model based on the route.
Models are doing too much. Sending emails, creating relationships, deleting and modifying other models, queuing tasks, etc. Basically now you have God-like objects that are supposed to do everything that may or may not concern with modeling and dealing with data.
Where do you draw the line? Isn't this just falling into the God pattern?
It might not be the best idea to look at Rails as a staple of MVC design pattern. Said framework was made with some inherent shortcomings (I kinda elaborated on it in a different post) and the community only just now has begun addressing the fallout. You could look at DataMapper2 development as the first major step.
Some theory
People giving that advice seem to be afflicted by a quite common misconception. So let me begin by clearing it up: Model, in modern MVC design pattern, is NOT a class or object. Model is a layer.
The core idea behind MVC pattern is Separation of Concerns and the first step in it is the division between presentation layer and model layers. Just like the presentation layer breaks down into controllers (instances, responsible for dealing with user input), views (instances, responsible for UI logic) and templates/layouts, so does the model layer.
The major parts that the model layer consists of are:
Domain Objects
Also known as domain entities, business objects, or model objects (I dislike that latter name because it just adds to the confusion). These structures are what people usually mistakenly call "models". They are responsible for containing business rules (all the math and validation for specific unit of domain logic).
Storage Abstractions:
Usually implemented using data mapper pattern (do not confuse with ORMs, which have abused this name). These instances usually are tasked with information storage-from and retrieval-into the domain objects. Each domain object can have several mappers, just like there are several forms of storage (DB, cache, session, cookies, /dev/null).
Services:
Structures responsible for application logic (that is, interaction between domain objects and interaction between domain objects and storage abstractions). They should act like the "interface" through which the presentation layer interacts with the model layer. This is usually what in Rails-like code ends up in the controllers.
There are also several structures that might be in the spaces between these groups: DAOs, units of work and repositories.
Oh ... and when we talk (in context of web) about a user that interacts with MVC application, it is not a human being. The "user" is actually your web browser.
So what about deities?
Instead of having some scary and monolithic model to work with, controllers should interact with services. You pass data from user input to a specific service (for example MailService or RecognitionService). This way the controller changes the state of model layer, but it is done by using a clear API and without messing with internal structures (which would cause a leaky abstraction).
Such changes can either cause some immediate reaction, or only affect the data that the view instance requests from model layer, or both.
Each service can interact with any number (though, it's usually only a handful) of domain object and storage abstractions. For example, the RecogitionService could not care less about storage abstractions for the articles.
Closing notes
This way you get an application that can be unit-tested at any level, has low coupling (if correctly implemented) and has clearly understandable architecture.
Though, keep in mind: MVC is not meant for small applications. If you are writing a guestbook page using MVC pattern, you are doing it wrong. This pattern is meant for enforcing law and order on large scale applications.
For people who are using PHP as primary language, this post might be relevant. It's a bit longer description of the model layer with a few snippets of code.
If the "model" classes are implemented poorly yes, your concern is relevant.
A model class shouldnt be doing Email (infrastructure tasks).
The real question is what does model in MVC imply.
It isnt restricted to POCO classes with a few methods.
Model in MVC means Data and Business logic. Treat it as a superset of classic core POCO models.
View ==== Controller ==== Model ---> Business Process layer --> Core models
Throw in Infrastructure assemblies and Data Access layers and use injection to hand that into the BPL then your a process is using MVC as intended.
BPL may invoke UoW / Respository patterns, and execute business rules and call Infrastructure features by way of injected Objects or interface patters.
So the recommendation to keep a controller skinny doesnt mean the "person" class in a classic Core model should have 50 methods, and call Email directly. You are right to think this is wrong.
The Controller May still be required to instantiate and inject Infrastructure classes into the BPL or core layer if called directly. There should be a business layer or at least classes orchestrating calls across Classic Object model classes.
Well thats my "view" anyway ;-)
For generic take on MVC the wiki description http://en.wikipedia.org/wiki/Model%E2%80%93view%E2%80%93controller
A Little Blog that talks about the "M" in MVC. http://www.thedeveloperday.com/skinny-controllers/
I think you can make a distinction between a single fat model (possibly named App or Application), and several fat models broken down into logical groups (Business, Customer, Order, Message). The latter is how I structure my apps, and each model roughly corresponds to a database table in a relational database or collection in a document database. These models handle all aspects of creating, updating, and manipulating the data that makes up the model, whether it is talking to the database or calling an API. The controller is very thinm responsible for little mor that calling the appropriate model and selecting a template.
Related
For a performance monitor I've implemented a MVC in Java and have decided to split up my Model into separate classes, now I'm torn between sorting its attributes and functions into Client, Server and Connection Model classes or Benchmark, Logging and Specs Model classes.
The former Models would match the real life counterparts as "objects", as they are the separate entities the application will monitor.
Due to how the MVC works with an observer, a benchmark function from my controller would trigger updates in all my Models.
The latter would solve this wide-spread updating by making Models match the functionality(requirements) more, thus only updating models that need to be updated, triggering only View-updates that need to be modified.
I think this approach would decrease my Model's re-usability though.
Classes should touch reality, but not reach it. If you were to stay on an absolute abstract level, your model would not be able to reach some kind of realization. When you model your domain you start with business objects that directly relate to (physical) objects in reality. Once you start designing a system you will derive a class model from the business model which is somehow linked. Usually you create dependencies from your system class design to the according business objects and you will also find a naming-parallel. The coupling between business and design model is loosely (dependencies) but existent. Each business object must have some trace to the design model. There will likely be more classes in the design model than in the business model.
Today I was getting some refresh about software design patterns.
In particular I was wondering about the difference between MVC and Three-tier-layer. Basically, from what I read on wikipedia and other sources, the main difference between the two is the components interaction:
A fundamental rule in a three tier architecture is the client tier
never communicates directly with the data tier;
whilst
...the MVC architecture is triangular: the view sends updates to
the controller, the controller updates the model, and the view gets
updated directly from the model
Now: if we take the apple docs regarding this matter we see this:
And they clearify that Views and Model shouldn't communicate directly:
view objects are typically decoupled from model objects in an MVC
application
and
When a model object changes (for example, new data is received over a
network connection), it notifies a controller object, which updates
the appropriate view objects
And so on.
So, what's the matter here? Is Cocoa adopting its own idea of an MVC, regardless of the common one? Or am I missing something in the common way of seeing an MVC architecture?
While it can be said that Cocoa's version of MVC is a sort of subset of the actual definition of MVC, they are not separate entities. The Cocoa version of MVC typically revolves around the use of a View (typically an NSWindow and/or an NSView), a controller (typically an NSWindowController), and a model layer (anything from a simple array to a Core Data stack). The separation of powers in this model is clear, but where in the 'tier' structure that Wiki defines should each of these belong?
I would argue that the Controller and the View are a part of the client layer. Not only is the controller responsible for the delegation between the model and the view, but it is responsible for responding to user events and determining the correct course of action to take during non-framework error handling. By taking this approach to MVC, you can now begin to see how Cocoa does, in fact, satisfy the broader definition of the pattern.
A fundamental rule in a three tier architecture is the client tier never communicates directly with the data tier;
This one's probably the hardest to reason about of the 3, and it involves delving into what "communication" actually means in the context of the pattern. When we say communication, what we mean is that the controller has no direct involvement in the actions taken by the model. That's not to say that the controller cannot order a change in the contents of the model, but rather that the controller does not have a hand in how the model updates itself. The controller acts as a director, not an implementer, which vastly simplifies the creation of a database layer, and is one of the reasons that Core Data and SQLite3 exist as external frameworks rather than as Foundation classes.
view objects are typically decoupled from model objects in an MVC application
That brings up one of the age-old taboos when programming with the pattern: making your views too smart. The controller provides a solid barrier between the model and view, such that the controller acts as a director and a filter for content from the model layer. Without any such barrer, say a tableview, would have to ensure that every cell had a copy of the data from the database, and that each cell knew when and how to update itself when a change is propagated in another cell. In Cocoa, this is where our NSWindowControllers come in. They manage the display of a root view and act as a barrier between some model and the content of the view it manages. Though, it is important to note that the controller objects in Cocoa are view-biased, mostly because it would be nearly impossible to provide a generic outlet to any kind of model layer without quite a bit of unnecessary glue.
When a model object changes (for example, new data is received over a network connection), it notifies a controller object, which updates the appropriate view objects.
That's the way it should be, for the reasons I've laid out above. But, to build on the networking example you've given, consider this:
Given an NSOperation that fetches data, and a controller that manages a tableview, you would probably not like the controller sticking its fat fingers into the operation, nor would you like the tableview to receive raw NSData and have to spend valuable rendering time processing the result and displaying it.
And so on. So, what's the matter here? Is Cocoa adopting its own idea of an MVC, regardless of the common one? Or am I missing something in the common way of seeing an MVC architecture?
I guess the conclusion I would draw from this is that your definition of the separation of powers in MVC and in how Cocoa does it is off. Cocoa is fairly rigid about adhering to the pattern, though there is an interesting contemporary movement within the Objective-C community towards MVVM.
You are correct the MVC practiced in most cocoa apps is not the MVC as it is defined in the text books. There are many variations of MVC employed by different frameworks. The MVC employed by tools with visual designers are heavily influenced by their visual designer implementation. With XCode you have story boards and nibs. The cocoa libraries and the way concerns are separated are influenced by this. If you want to take advantage of these tools, I would recommend understanding how concerns are separated by Xcode and work within this approach. Your code will coexist with it more smoothly. Apple documentation will help with this.
That being said, MVC is about separation of concerns. Separating concerns is hugely important in designing and maintaining software. Separating concerns properly can reduce dependency, reduce cyclomatic complexity, and make your code much more testable and maintainable. I think it is good that you are paying attention to this and whatever way you structure MVC should look to the reason why you are separating concerns as the guide to implementation.
I just cannot seem to get my head around what exactly is the MODEL in MVP.
If I have a layered architecture PRESENTATION / APPLICATION / DOMAIN / INFRASTRUCTURE, what exactly is the MODEL?
DOMAIN objects accessed through
lower layers?
A separate object defined in the
PRESENTATION layer that maps to the
UI and uses data obtained from a
lower layer?
If someone could clear my understanding on what is the MODEL it would be greatly appreciated.
The Model is normally the group of classes/types/components that represent the core domain (business or otherwise) that your application operates within. These are the classes that perform the key logic required, often in the form of business rules, and also consume/manipulate data.
In your layered example, the Model would mostly be found in the Domain layer but could also be in the Application layer.
I think you're having difficulty understanding it because you are trying to combine two separate architectural patterns, or ways of looking at the application, being n-tier/n-layer versus MVP.
It's completely reasonable (and quite common) to use some sort of Model/View approach while at the same time applying layering in your application.
Maybe you should focus on them one at a time to start with and then overlay them when you are more familiar with both.
In any of the Model-View-* architectures, the Model is what describes the data in your application (and, if they fit the need, are passed in to the View for rendering).
If your application already has Domain objects, it very well may be the case that you could use them for your Model.
It doesn't matter what architectural guidelines you're following, M is always going to be the same thing. The Model is the piece that is specific to your domain. It's the part that really is what you're application is trying to do. The Model is supposed to represent your business domain. This goes for MVP, MVC, MVVM, etc.
If you were making a inventory system, then an Inventory class would most likely be in your Model, a Product would probably be there, an Order, you get the idea. These are the things that compose your domain logic.
The model is the data. This might just be data out of a database in DataSets, or it might be a complete domain model with objects representing your field of business.
The view is the UI, whether web pages or a Windows application or a mobile device application.
The presenter is the glue between the two and the brains of the whole outfit. Actions initiated by the view take place in the presenter. Generally in a WinForms application, for instance, a Button.Click event in my View simply calls a method on the Presenter, which then takes whatever action is necessary (and it may just be doing something back in the View).
The presenter holds a reference to the view (through an interface), and to the model. The view has a reference to the presenter (usually I strongly-type this, but it can be an interface as well). The model doesn't know about the presenter or the view.
I am new to the concept of MVC and multi-tiered web architecture. I developing a PHP application and am using one of the available MVC frameworks. My question is as follows:
From what I understand, MVC in and of itself is not considered a multi-tier architecture. I can understand how using MVC alone is a step up from taking an unstructured approach, but I was contemplating how a simple 3-tier architecture would fit in? Would MVC reside in the presentation layer? What are the merits of adding a tiered approach? From what I gather, with MVC alone there are no explicit data objects responsible for retrieving data from the database and this is usually stuffed into the model. Likewise the business logic, which in a 3-tiered architecture would reside in a 'business-layer' (or whatever you want to call it), can be stuffed into the controller.
Is my understanding somewhat correct? I know I asked a lot of questions, but I would like to hear you discuss how you incorporated an n-tier architecture into your MVC framework (PHP or otherwise) as I assume that the two are not mutually exclusive. Thanks!
M) M is your model. This is generally living in your business layer or the layer just behind your presentation layer. Many people don't like the presentation layer to have any knowledge of the business layer though and so they further abstract that by having what is called a ViewModel. These frequently are DTO (data transfer objects) that loosely map to your Domain model. For me (.net guy) there are tools such as AutoMapper to make the conversion from Domain Model to View Model.
V) V is your view. The view is your presentation layer. This is the actual HTML or PHP code that the user directly touches and interacts with. The view should be as light as possible (meaning no logic if possible). Try to keep any sort of if/then type scenarios out of the view and stick to just displaying and collecting data. Present a ViewModel to your web designers so that they don't contaminate your DomainModel.
C) C is your controller. This is much like a co-ordinator. It takes data from your view and makes sure it gets to the right back end function/method for processing that data. It also co-ordinates data from the back end on it's way to the front end.
Where multi-tier design concepts come in is behind the Presentation layer (where is where MVC is primarily). When the controller is taking data from the view and passing it back to the back end it (if you follow DDD or Domain Driven Design) would pass the data to an application service (a class that co-ordinates back end matters). The service might further push the data into a Repository layer (which is a class that speaks to the database, filesystem, web services, etc. - any infrastructure stuff). DDD is a big topic but will get your head around the n-tier approach and how it works with MVC.
While researching this topic take a look at IoC (inversion of control), DI (dependency injection), TDD (test driven development), and as many patterns as possible (facade, factory, etc.).
In general, business logic should not be in the Controller - you'd end up with massive controllers if you followed this pattern. The Model basically houses all your non-presentation layers ... data access, business logic, and business entity objects. Your controller prepares the business data for the view.
In many leading DDD projects, especially MVC style, I see the UI using display objects that mirror domain entities, rather than using those domain objects directly. This style is obviously for decoupling and separation of concerns, and I personally prefer this style.
But what I'm not sure of, is whether this a strict tenet of DDD, or whether this is more just different developers' interpretation of it.
Can you use your domain objects directly in the UI, and still be following the DDD methodology in that act?
Or is it a DDD best practice to always use display objects?
Note: While I mention MVC, I'm really interested in whether display objects must be used in almost all DDD compatible UI patterns in a DDD project.
I didn't really start to understand why or how you would decouple the domain model from presentation concerns until I started following the work of Greg Young and Udi Dahan (via Martin Fowler).
They have been teaching a principle known as Command and Query Responsibility Segregation (CQRS).
My interpretation of CQRS is that there are two sets of responsibilities that can pull a domain model in different directions, resulting in a model that does a mediocre job of both. The two responsibilities are commands (i.e. behavior that would cause a write to the database) and queries (i.e. reading from the database in order to fetch data for UI consumption). An example would be adding getters and setters to your entities to support databinding in the UI. If your model has getters and setters, it will probably do a poor job of modeling state changes that need to happen transactionally or encapsulating business logic. It will also have no way of modeling the business context of state changes (see Event Sourcing).
In DDD terms, you might say that the domain model and the presentation model are usually in separate Bounded Contexts.
The solution as prescribed by CQRS is to create one model for commands and another for queries. If your current model has methods for changing state (i.e. the behavior in the model), and getters that expose state to a UI for databinding, you would refactor these two responsibilities into separate models for commands and queries. The query model will not be mapped to your domain entities, but instead directly to the database. If your database doesn't capture derived state from your domain model, check out a pattern called Eager Read Derivation.
If your system is simply CRUD and has no behavior, try out a scaffolding system that can be automatically built off of your database schema, like ASP.NET Dynamic Data
If you're doing an MVC pattern, you need view objects; the DDD is just your model. That doesn't mean you must always use MVC; a DDD could be built, say, as a simulator, where all you look at is log messages emitted. But in MVC you really should have separate view objects.
Now, ask yourself why that would be? The answer is that the view can change, even though the business model doesn't. The DDD model should express, in the business's terms, what is essential to the business.
DDD is a way of thinking while designing a software that starts with modelling the domain. As the webpage puts it:
Domain-driven design is not a
technology or a methodology. It is a
way of thinking and a set of
priorities, aimed at accelerating
software projects that have to deal
with complicated domains.
One thing that follows naturally out of this design pattern is a layered architecture. As it is said in DDD Pattern Summaries
Partition a complex program into
LAYERS. Develop a design within each
LAYER that is cohesive and that
depends only on the layers below.
Follow standard architectural patterns
to provide loose coupling to the
layers above. Concentrate all the code
related to the domain model in one
layer and isolate it from the user
interface, application, and
infrastructure code. The domain
objects, free of the responsibility of
displaying themselves, storing
themselves, managing application
tasks, and so forth, can be focused on
expressing the domain model. This
allows a model to evolve to be rich
enough and clear enough to capture
essential business knowledge and put
it to work.
Whether you need to have display objects to do that? That is just one way of implementing this, but might not even be the best to provide loose coupling. As for an example: maybe the view layer is but a webbrowser and xlt files to visualize xml files emmitted by the business layer? If anybody has more fitting examples, please add them. My point is that DDD stresses a layered architecture, but does not limit this to one possible solution.
Within an MVC design you would typically have a mapping from Repository -> Domain Models and then from Domain Models -> View Models. The View Models often contain Domain Objects though.
The answer is quite straight :
The domain objects have domain oriented design and methods.
The view objects have view/control oriented design and methods.
If you can use your domain objects in the view, they are maybe not quite domain oriented.
Domain objects are really the internal logic inside your business logic layer. They shouldn't be directly exposed to your clients (UI layer). Instead, encapsulate the usage of your domain model into application services. The application services can utilize lightweight DTOs and/or command objects to pass data and intent between the client and the domain model.