In an ASP.NET MVC project we are using AutoMapper to map from domain model to viewmodel - and sometimes also flattening a hierarchy while doing so. This works like a charm and makes the rendering logic of our views very lean and simple.
The confusion starts when we want to go the other way from viewmodel (or postmodel or editmodel) to domain model, especially when updating objects. We can't use automated/two-way mapping because:
we would have to unflat the flattened hierarchy
all properties on the domain model would have to be mutable/have public setters
the changes coming from the view isn't always just flat properties being mapped back to the domain, but sometimes need to call methods like "ChangeManagerForEmployee()" or similar.
This is also described in Jimmy Bogards article: The case for two-way
mapping in AutoMapper, but the solution to this isn't described in detail, only that they go:
From EditModel to CommandMessages – going from the loosely-typed
EditModel to strongly-typed, broken out messages. A single EditModel
might generate a half-dozen messages.
In a similar SO question there is an answer by Mark Seeman where he mentions that
We use abstract mappers and services to map a PostModel to a Domain Object
but the details - the conceptual and technical implementation - is left out.
Our idea right now is to:
Recieve a FormCollection in the controller's action method
Get the original domain model and flatten it to viewModelOriginal and viewModelUpdated
Merging the FormCollection into viewModelUpdated using UpdateModel()
Use some generic helper method to compare viewModelOriginal with viewModelUpdated
Either A) Generate CommandMessages a la Jimmy Bogard or B) Mutate the differences directly into the domain model through properties and methods (maybe mapping the 1-1 properties directly through AutoMapper)
Can someone provide some examples of how they come from FormCollection through editmodel/postmodel to domain model? "CommandMessages" or "abstract mappers and services"?
I use the following pattern:
[HttpPost]
public ActionResult Update(UpdateProductViewModel viewModel)
{
// fetch the domain model that we want to update
Product product = repository.Get(viewModel.Id);
// Use AutoMapper to update only the properties of this domain model
// that are also part of the view model and leave the other properties unchanged
AutoMapper.Map<UpdateProductViewModel, Product>(viewModel, product);
// Pass the domain model with updated properties to the DAL
repository.Update(product);
return RedirectToAction("Success");
}
You might want to consider CQRS(Command Query Responsibility Segregation - I think this might be the concept you were missing), possibly even with Event Sourcing.
It is basically a practice of separating the logic of reading from a data source and writing to a data source, might even mean having different data models for reading and writing.
This might be a good place to start: http://abdullin.com/cqrs/
Option C: Put it all in the controller action. Next, if that gets hairy, decompose into services (abstract mappers) or messages-as-methods (the command message way).
Command message way:
public ActionResult Save(FooSaveModel model) {
MessageBroker.Process(model);
return RedirectToAction("List");
}
And the processor:
public class FooSaveModelProcessor : IMessageHandler<FooSaveModel> {
public void Process(FooSaveModel message) {
// Message handling logic here
}
}
This is really just about moving the "processing" of the form out of the controller action and into individual, specialized handlers.
But, I'd only really go this route if controller actions get hairy. Otherwise, just take the form and do the appropriate updates against the domain models as necessary.
There are some similarities here with what I've been doing. My hierarchy of view models is only somewhat flattened from its domain object equivalents, but I do have to deal with calling explicit service methods on save to do things like adding to child collections, changing important values etc, rather than simply reverse mapping. I also have to compare before and after snapshots.
My save is Ajax posted as JSON to an MVC action and enters that action magically bound back to a view model structure by MVC. I then use AutoMapper to transform the top level view model and its descendants back into its equivalent domain structure. I have defined a number of custom AutoMapper ITypeConverters for those cases where a new child item has been added on the client (I'm using Knockout.js) and I need to call an explicit service method. Something like:
foreach (ChildViewModel childVM in viewModel.Children)
{
ChildDomainObject childDO = domainObject.Children.Where(cdo => cdo.ID.Equals(childVM.ID))).SingleOrDefault();
if (childDO != null)
{
Mapper.Map<ChildViewModel, ChildDomainObject>(childVM, childDO);
}
else
{
MyService.CreateChildDO(someData, domainObject); // Supplying parent
}
}
I do a similar thing for deletes and this process cascades quite nicely down through the whole structure. I guess a flattened structure could be either easier to work with or harder - I have an AbstractDomainViewModel with an ID with which I do the above matching, which helps.
I need to do comparisons before and after updating because my service layer calls trigger validation which can affect other parts of the object graph, and this dictates what JSON I need to return as the Ajax response. I only care about changes which are relevant to the UI, so I transform the saved domain object back to a new view model and then have a helper method to compare the 2 view models, using a combination of manual upfront checks and reflection.
Related
I am new to MVC and am having a conceptual problem with state and object persistence and hope someone can put my thoughts in order.
I have a remote webservice which provides methods to manage orders. An order consists of a header and Lines as you would expect. Lines can have additional requirements.
I have my domain objects created (using xsd2code from the webservice schema), the webservice calls and object serialization all working fine. I've build the DAL/BLL layers and it's all working - tested using a WinForms testbed app front-end.
I have view model objects mapped from the domain objects using Automapper. As the order is returned from a single webservice method complete with lines etc I have an OrderViewModel as follows
public class OrderViewModel
{
public OrderHeaderViewModel OrderHeader { set; get; }
public List<OrderLineViewModel> OrderLines { set; get; }
public List<OrderLineAdditionalViewModel> OrderLineAdditional { set; get; }
public List<OrderJustificationViewModel> OrderJustifications { set; get; }
}
Firstly I'm wondering if I should dispense with the OrderViewModel as if I pass this as a model to a view I'm passing far more data than I need. Views only need OrderHeader or OrderLines etc - not the entire order.
Now my conceptual problem is in the controllers and the views and object persistence.
My Order controller has a Detail Action which performs the load of the order from the webservice and maps the Domain object to the OrderViewModel object.
public ActionResult Details(string orderNumber)
{
OrderViewModel viewModel = new OrderViewModel();
var order = WebServiceAccess.LoadOrderByOrderNumber(orderNumber,"OBOS_WS");
viewModel = AutoMapper.Mapper.Map<BusinessEntities.Order, ViewModels.OrderViewModel>(order);
return View(viewModel);
}
But the Order/Details.cshtml just has the page layout and a call to two partial pages for the header and the lines (I swap the Headerview for a HeaderEdit using Ajax, same for the LinesView)
#{ Html.RenderPartial("DetailsHeaderViewPartial", Model);}
#{ Html.RenderPartial("DetailsLinesViewPartial", Model);}
At the moment I'm passing the model into the main Details container page, then into the RenderPartials, However I don't think that the model should be passed to the main Detail page, as it doesn't need it - the model is only needed in the DetailsHeaderViewPartial, DetailsLinesViewPartial so I'd be better off using #RenderAction here instead and passing the model into the Header/Lines views instead.
However, The Order is retrieved from the webservice in the ActionResult Details() how can I make the retrieved OrderViewModel object available in the ActionResult HeaderDetails() / LineDetails() methods of the controller to pass as the model in return PartialView(...,model) ?
Should I use a User Session to store the Order ViewModel so it can be used across actions in the controller.
Moving on from this stage the user will be able to maintain the order (add/remove lines - edit the header etc). As the webservice call to save the order could take a few seconds to complete I'd rather only call the save method when the user has finished with the order. I therefore would like to persist the in-progress order locally somewhere whilst it's being worked on. User session ?
Many thanks for any advice. Once I've got my head around state management for the ViewModels I'll be able to stop reading a million Blog posts and actually write this thing !
You actually have a few questions here so I will try to address them all the best I can.
1) Dispensing with the view model : I would say no. The view model represents the data that you need in order to populate your view. It seems like you are using the view model as an identical container to the domain model object. So you are asking if you should dispense with it and just pass the domain model to the view while your original concern is that you are passing along more data then you really need as is?
Rather then dispensing with the view model, I would revisit your properties on your view model. Only use properties that you need and create the mapping logic (either with automapper or on your own) for taking the complex domain object and populating the properties on the view model.
summation: build the view model to be only things that the view needs and write mapping logic to populate that view model.
2) This is just a statement of best practice before I breakdown your specific scenario.
You describe your architecture as having a BLL and DAL. If that is the case then you should not be persisting any objects from your controller. The controller should not have any knowledge of the database even existing and the objects used in the controller should have no idea of how to persist themselves. The objects that are going between your controller and the web service should strictly be Data Transfer Objects (DTO's). If you are unfamiliar with what constitutes a DTO then I highly suggest that you do some research and try to build them into your solution. It will help you conceptually see the difference between view model objects, domain objects and data transfer objects.
3) I would not try to store an order object in the session. I would re-analyze how you are breaking up the partial views within the view so that you can call actions with the ordersviewmodel being the parameter in a way that you need. It sounds like you are needlessly breaking up views into partial views.
4) You should not be concerned with state management for the view model object. Your view (which can be comprised of many partial views) is filled based on properties provided by the view model. The user can make changes using the UI you have developed. Since you express the desire to only save once they are finished making all changes to optimize calls to the web service, you just need to repopulate the fields of the view model upon clicking submit. Now you have a "state" for orderviewmodel that represents the users changes. You can send this object to the web service after converting back to a DTO (if you do what I said above) or by mapping it to the domain object.
1 final note. You are using automapper to map your domain to the view model. I am assuming that your view model is too complex and includes things that you don't need because you built your view model to emulate the domain object so that automapper could map by naming convention. Automapper has an api for doing complex (custom) mappings that fall outside of standard same name properties. Don't let automapper constrain you to building your view models a certain way.
Hope this helps
want to get rid of ViewBag in my views so i've made a little research about viewmodels.
i like the idea of using it for presentation, no questions here.
but i don't realise what is the best way to use viewmodels for update.
first of all, why shouldn't i use my EF entities with [MetadataType(typeof(User_Validation))]?
public ActionResult Edit(User user)
{
...
}
where User is EntityObject.
then, if i however use viewmodels for it there is another question: here people believe that nested viewmodels should be used, but here is another opinion ("They are not wrappers around domain models", he says). Who is right?
also, what is the best way to update an object in action after POST (without using tryupdatemodel, because here and not only, people are against this approach). i've tried to use ApplyCurrentValues but if there is some complicated logic of update with many-to-many relations, for example, i get some huge EF errors. that's why i need to manually set fields of my EntityObject that i get from DB with values that come to action. Smth like this:
public ActionResult Edit(User user)
{
if (ModelState.IsValid)
{
var userToUpdate = usersRepository.Get(user.UserId);
userToUpdate.Field1 = user.Field1;
...
// save
}
...
}
where User is EntityObject with validation class.
so, if i use flat viewmodels to get values from form and AutoMapper to set values to my entityobject will it be the most right way to deal with updates or it can be automatized even more?
You answered yourself in the last paragraph. I am using the exactly same approach in my MVC3 EF Code First project.
Here is the practice I follow:
All entity classes are clubbed under one folder "Entity". You can also opt for a separate library project.
Every entity has a corresponding ViewModel class with a postfix of Model (e.g. for Profile entity there is ProfileModel class).
For nested entities there are corresponding nested ViewModel classes.
AutoMapper is used for Model to Entity and vice versa conversions. Here automapper takes care of the nested entities. In case of complex updates, rather than relying on AutoMapper, I take the matter in my own hands in controller.
Works flawlessly with healthy separation between my domain objects and view models.
I am pretty new to MVC and I am looking for a way to design my models.
I have the MVC web site project and another class library that takes care of data access and constructing the business objects.
If I have in that assembly a class named Project that is a business object and I need to display all projects in a view ... should I make another model class Project? In this case the classes will be identical.
Do I gain something from doing a new model class?
I don't like having in views references to objects from another dll ... but i don't like duplicating the code neither.
Did you encounter the same problem?
I usually use a combination of existing and custom classes for models, depending on how closely the views map to the actual classes in the datastore. For pure crud pages, using the data class as the model is perfect. In real world use though you usually end up needing a few other bits of data. I find the easiest way to combine them is something like
// the class from the data store
public class MyDataObject {
public string Property1 {get;set;}
public string Property2 {get;set;}
}
// model class in the MVC Models folder
public class MyDataObjectModel {
public MyDataObject MyDataObject {get;set;}
// list of values to populate a dropdown for Property2
public List<string> Property2Values {get;set;}
// related value fetched from a different data store
public string Property3 {get;set;}
}
I also have quite a few model classes that are just a set of form fields - basically corresponding to action parameters from a particular form rather than anything that actually ends up in the database.
If it is exactly the same project then obviously don't need to duplicate the Project class, just use it as is in the view. But in real life often views have specific requirements and it it a good practice to define view model classes in your MVC application. The controller will then map between the model class and the view model to be passed to the view.
You will likely find differing opinions on this. I will give you mine:
I tend to, by default, reuse the object. If the needs of the view change and it no longer needs most/all of the data in the business object, then I'll create a separate view. I would never change the business object to suit the view itself.
If you need most/all of the information in the business object, and need additional data, then I would create a view that holds a reference to the business object and also has properties for the additional data points you need.
One benefit of reusing the business object is that, depending on the data access technology you are using, you can get reuse out of validation. For isntance, ASP.NET MVC 3 coupled with Entity Framework is nice as they both use the attributes in the System.ComponentModel namespace for validation.
That depends on what you mean by model. In the architecture I typically use, I have my Domain model, which is a collection of classes in a separate class library. I use DataAnnotations and built in validation to automatically validate my model when saving.
Then there is the View model, which I place in my MVC project. View models only have the information relevant to the view. I use data annotations here as well since MVC will automatically use them for my validation.
The reason for splitting the model this way is that you don't always use every piece of your domain model in a view. That means you either have to rebuild the data on the backend or you have to stash it in hidden fields. Neither is something I like,.
Here is the lay of the land. Like most people I have my domain object and I have my view models. I love the idea of using view models, as it allows for models to be created specifically for a given view context, without needing to alter my business objects.
The problem I have is with type level validation defined on my domain object and getting those rules to the client. In this case lets say I am using data annotations to describe the validation rules, when I move the data from the domain object to the view model, the view model no longer knows what validation it should get the interface to perform (because the validation is defined back on the domain object).
With MVC 2 you can get it to automatically perform client/server side validation, based on the validation rules of the current object. But because the validation rules are defined on the domain object and not the view model, I would have to duplicate the validation rules on the view model to get this to work.
How do others deal with this type of issue? My thinking is that besides mapping the data from the domain object to the view model, we also need to map across the validation rules, but I haven't really seen others talking about this issue... Brad Wilson has recently talked about this issue at length but hasn't really addressed the duplication of rules on the domain object and on the view models... what are your thoughts?
Cheers
Anthony
The DataAnnotation attributes are about validating input and giving UI feedback to the end user. That's really their only intended use. I use different validation strategies for UI objects and business objects, so the DA validation attributes only end up on models being shown to the user.
This may not be appropriate, but what if you just moved your validation rules/annotations from your Models to your ViewModels? In a few of the projects I've been on, we've chosen to prevent the View from accessing anything but information exposed through its corresponding ViewModel. Since all data interaction would be performed through the ViewModel, there wouldn't be a need to have validation on your Model objects.
The counter to this argument is that you could easily duplicate certain validation rules, since different ViewModels might be interfacing with the same Models. In this case, it might make sense to simply declare your Model as a property exposed on your ViewModel. For postbacks, they could accept a Model as their parameter, allowing the ModelBinder infrastructure to handle the request. In this case, if ModelState.IsValid is false, you could just reassign the property to your ViewModel before redisplaying the View.
I would recommend moving your annotations to your ViewModels. It makes sense since a lot of Views are a) the result of composition of several models or b) a subset of model data.
It turns out that AutoMapper may be able to do this for us automagically, which is the best case scenario.
AutoMapper-users: Transfer validation attributes to the viewmodel?
http://groups.google.com/group/automapper-users/browse_thread/thread/efa1d551e498311c/db4e7f6c93a77302?lnk=gst&q=validation#db4e7f6c93a77302
I haven't got around to trying out the proposed solutions there, but intend to shortly.
(Cross posted this on my (dupe) question as well).
Probably we shouldn't use view models at all?
And define validation rules on model layer entities..
I've been considering this as well for a while now. I totally understand Brad's reply. However, let's assume I want to use another validation framework that is suitable for annotating both domain entities and view models.
The only solution I can come up with on paper that still works with attributes would be to create another attribute that "points" to a domain entity's property that you are mirroring in your view model. Here's an example:
// In UI as a view model.
public class UserRegistration {
[ValidationDependency<Person>(x => x.FirstName)]
public string FirstName { get; set; }
[ValidationDependency<Person>(x => x.LastName)]
public string LastName { get; set; }
[ValidationDependency<Membership>(x => x.Username)]
public string Username { get; set; }
[ValidationDependency<Membership>(x => x.Password)]
public string Password { get; set; }
}
A framework like xVal could possibly be extended to handle this new attribute and run the validation attributes on the dependency class' property, but with your view model's property value. I just haven't had time to flesh this out more.
Any thoughts?
I have been learning ASP.NET MVC for a few months. I have learned about views and controllers and models and stuff. To design a view, we always need a model.
Usually a model is a just a class which we fill with data and pass to a view. I have a question here: Should a model itself do some calculation, or should it just be dumb?
For example, I have a site where I load Books by Users. My model class is as follows:
public class FormViewModel
{
public User MyUser {get; set;}
public Books UserBooks {get; set;}
//Constructor here.
public FormViewModel(User _user, Books _userBooks)
{
this.MyUser=_user;
this.UserBooks=_userBooks;
}
}
This class doesn’t do anything — it’s just a template. Now if I modify the code as follows:
public class FormViewModel
{
public User MyUser {get; set;}
public Books UserBooks {get; set;}
//Constructor here.
public FormViewModel(User _user)
{
this.MyUser=_user;
this.UserBooks=_user.GetBooks();
}
}
which Books are collected depends on which User has been selected. Now it’s much easier to work with.
I just want to know what is a good approach according to MVC patterns and practices.
You want to separate all of your business logic, and data validation into the model. Usually that includes "grouping" datasets and such or filtering data by some criteria.
You want to separate all calls to these methods of the model in the controller, who's responsibility is to retrieve and send data to and from the model. The controller then passes the applicable data set into the view.
The Helpers are logic that is used by the view to do presentation logic (not business logic or validation) such as printing menus and such.
The View is where you will use the Helpers (or not, they are not required to use MVC properly but they do "help" :p) to write HTML, CSS and JS to the browser. You may also separate commonly used view modules into partial views that you can include on more then one view.
You can further seperate things into a ViewModel, but then you are going outside of "strict" MVC. In this case, you would use the ViewModel to help the view interact with the model - basically the ViewModel is a modular controller. In this case, the controller would do much less then what little it does already.
However, this is generally overkill for web applications. Because web apps have a single flow of execution (the request) separating things into a ViewModel becomes unnecessary. However, in GUI code, the ViewModel becomes much more useful (as GUIs have much more then a single flow of execution).
You always want to separate business logic into the Model, period. Remember that you should not couple your controller to your model - so that you can use your model elsewhere in other controllers or even expose it as a web service.
Hope this helps
:)
You can do it a couple of ways, but I'd say the easiest way would be to pass in the reference identifier for the user which you are trying to access through to the controller action (as below) and let it do all the data access calls.
public void GetUserAndDetails(Guid userId) { ... }
Then in this controller action you can look up the user details and the books for this user, set the properties on the view model instance and return it to the view to access.
FormViewModel model = new FormViewModel();
model.MyUser = GetUser(userId);
model.UserBooks = GetUserBooks(userId);
return View(model);
This way the view remains dumb (which it should be) and the model is relatively simple. This also helps with testing purposes.
Hope this helps.
In general, this kind of work should be done in the model. This is for a couple of reasons. First, if getting the user books requires a database connection, you don't want to be doing that from the view - it will just slow it down. The other thing to remember is that there could be multiple views, and you would need to duplicate that code in all of the views (web clients, maybe rich clients, etc).
In the MVC pattern, the views should be the "dumb" pieces. This allows easier use of multiple views and changing views when needed. It's also easier to test code when it doesn't require a view, so you could test the model without bringing up a web client.
Jeff
it is the view that is "dumb". all it does is display the manipulated data.
the controller simply fetches data from the model for the view... again the controller is ONLY a go between.
the model does everything. it stores the data and contains the classes and methods that manipulate it.
When you have a viewmodel that is different from your domain model, you shouldn't map the domain model to the viewmodel inside the viewmodel. It would make the viewmodel class responsible for more than one thing. You can do the mapping in the controller or in a service layer.
Check out this. You are talking about VIEW model, not domain model. There is a huge difference. View model should be dumb, it's just a placeholder for data which allows your views to be more strongly typed. Domain model must be a heart of your app, it must contain ALL business logic.
MVC is a pattern by itself. I really don't have any experience with ASP.NET MVC, but worked several times and used MVC pattern. Sometimes, MVC is realted to other development patterns like heartbeat, memento, state ...
No business logic is allowed for a Model! It's a bad design!
Your logic must be in the Controllers and more exactly: place your logic into the Helpers (the Helpers might consume your BLL and/or DAL) and then use your Helpers in your Controllers.