We have been using ExtJs for a long time, and now, with the advent of MVC pattern in it we are facing a critical question of whether to use it or not in the times ahead given the approach we had been following so far.
In our application, we have similar screens, like 40 similar screens for generating the reports and so on. These report screens would be having a form with buttons - PDF, CSV, EXCEL. All the buttons would be having same functionality at every screen, that is, submit the form to a url and generate the specific report.
The approach which we had been following was to create a custom class, say reportscreen class, which extends the window class of ExtJs.
This custom report screen class would already have buttons with their handlers defined (given the fact that the functionality is similar for them for all the screens) and will also have a form present with a function setFormItems(). Any instance of this class would provide the items for the form using setFormItems() and the url for generating the report.
This approach not only helps us in reducing the development effort to a great extent, but also allows even a new member of the team to use the class & create a new screen.
Now comes in MVC, the most advised pattern, and the approach shared above is definitely not as per MVC pattern. But then, do we need to really move to MVC pattern here?
Using the above approach, we are not only able to save the effort, but also have a greater control over things.
The purpose of asking this questions is to know about the exact advantages we shall get if we implement the above scenario using MVC pattern and move away from our current approach.
Thus, what extra things would MVC bring to the table in the context above? And also, what would be the best way to implement such a thing in MVC pattern?
Thanks for any help in advance.
Do you need to use MVC? No. Assuming you've already got a structure that works well for you, if it's not broke, don't fix it. If you were starting a new project I'd recommend using MVC, but I wouldn't refactor a whole bunch of code for the sake of it. If you did use MVC, you would follow a similar approach. You'd have a base view class where you dynamically push on the items in the subclass. You'd have a base controller to handle all the custom events, then subclass controllers to implement custom functionality on a "per module" basis.
It's hard to explain the advantages of using MVC to someone who had no experience with it. As someone who developed with both ExtJs 3 and 4 (with MVC), I think one of the greatest advantages of MVC is that it forces you to think right - whether you want it or not, using MVC is likely to result in code base that is more reusable and better encapsulated. But if you boys and girls are good programmers/designers, this will be the case anyway.
If you do choose to move to MVC, I believe that the main change will be in that now instead of handling user interaction within your custom class, you will have a single controller to do the same job - leaving your custom class to realize the view only. Controllers in ExtJs4 are global - meaning that the same controller may control 40 similar views. So no duplication of controllers here.
It's important to mention that even within an MVC design you will find non-MVC code. For example, I have a custom record editor which is a form - the component knows how to handle user actions and update the records accordingly (the form is not submitted as it's mostly dealing with associations). This component takes the equivalent role of both the controller and the view within an MVC design. It seems to me more than reasonable to encapsulate things this way, rather than split this to a view and a controller. But the (non-reusable) user screens are implemented using MVC. So maybe just another reason not to rush to MVC.
While I agree with Evan that if it aint broken don't fix it, had I been your development leader and given time and money is not an issue, I would actually request the gradual migration to MVC - overall I think that 3 years down the line you won't regret this decision. I would at least attempt it.
When ExtJs 4 came out I scraped 2 months of work to rewrite from scratch a system using MVC - I don't regret this for a second now (I did at the beginning, but not for MVC - rather for the amount of bugs in the early versions of 4).
Related
I'm scratching the surface with Angularjs, and thought I'd run a conceptual question past the fine folks of SO. This is a newbie question from a seasoned developer.
The application has dashboard requirements... a single page that surfaces stuff from many parts of the application. Different user types get different dashboards. We already have a legacy back end, so the first task is to build the dashboard to show many bits from it's new RESTful service layer.
I'm wondering how I should conceptually think about the controllers needed to support this?
The first question is... should they be model-centric or view-centric? In other words, should they be "view-centric" controllers that have the word "Dashboard" in them? Or should they be more focused on the model elements they represent, like "Tasks", "Contacts", "Notifications". Or should there be both where the dashboard controllers work with model-centric controllers?
The next question is... what level of granularity should the controllers represent? If view-centric "Dashboards" controllers, should they be "ManagerDashboardController" and "WorkerDashboardController"? If model-centric controllers, should there be controllers such as "LateTasks" & "OnTimeTasks" since I need to display them on different sections of the dashboard, with slightly different data?
I'm looking for tangible advice based on real-world experience and/or references to great links I've yet to find.
Here are my views from developing business applications in Angular for the past 6 months:
Role of a Controller
Initialization (loading initial data, setting options)
Exposing variables and functions to the template through the $scope
Application flow (through exposure of functions that can change state, or $watches)
I have found that, much like in traditional MVC frameworks, the controllers in an Angular app should really be super slim. Little if any business logic should be in the controllers, and should be instead be encapsulated in your models. I came to this conclusion after hearing the follow line from a presentation by Miško Hevery: "the purpose of the scope is to refer to the model and not be the model." That was the most valuable and enlightening line I got from that presentation (though I recommend to watch the whole video); that line directly resulted in me slimming down my controllers by almost 50%-70%.
For example, my company has a concept of an Order. I created a model that encapsulated all the properties of this business object, as well as its behaviours and injected them into the controllers. One business rule we had was the ability to add a Booking (another business object) to the Order. Originally in my controller, I had a $scope.addBooking function that first created a new Booking, then took the order and did a $scope.order.bookings.push(newBooking). Instead, I moved this business logic (addBooking function) directly into my Order model, and in the template I could then do something like <button ng-click="order.addBooking()">Add Booking</button> without adding a single line of code into my controller.
A lot of the code I put in my controllers when I was first starting off with angular, I found could be stripped out and placed either in my models, directives, or services (mostly the first two in my case). The remainder of the code left in my controllers almost all fell into one of the above 3 roles I listed above. It was either initialization code (e.g. firing an AJAX request to fetch data of the relevant business objects), scope assignment of objects, or scope assignment of functions that dealt with application flow (e.g. $scope.save or $scope.cancel that might send you back to a different page).
Should controllers be model-centric or view-centric?
This is an interesting question, one that I haven't thought about before. When I hear view-centric, I think of a controller that deals primarily with the view and how things are displayed. I feel there shouldn't be any controllers that are purely view-centric, the reason being it seems a view-centric controller can probably be transformed into a directive. You mentioned view-centric controllers as being like a Dashboard controller, which sounds like something that could definitely be made into a generic directive. Your directives should encapsulate most of your view logic, while your controllers focus on simply exposing your models to the view for consumption (back to the role of the controller). This has me thinking that controllers should more often be model-centric.
I think really the only conclusion I can come to is if a controller starts becoming too view-centric (with many variables and functions that deal primarily with the view and UI behaviour) then that is a sign that parts of your controller can be pulled out into a directive, making your controller slimmer.
This is very subjective but here is my answer to your questions
should controllers be model-centric or view-centric?
It depends (as always), I always try to have small controllers for the different parts of the page.
Some parts of the page are very view-centric (typically the ones that are shared among the different views). I usually have a menuCtrl, a headerCtrl and footerCtrl. This ctrls are very coupled to those parts of the page so a make them view-centric.
The other parts of the view, the ones that are business related are much more coupled to the business rules and in extension to the model so I make those ctrls model-centric. On an account´s business app, I would probably have an accountCtrl, an ownerCtrl, and so on. By doing so I can reuse them on different views if needed (and are much easier to test)
what level of granularity should the controllers represent?
The smallest as possible. Try to have small controllers that prepare the model for different parts of the page. If you have a big controller it will be hard to test, maintain and you will probably be forced to duplicate code on different parts of your application.
advices and recomentations with controllers
Keep them small.
Avoid DOM manipulation inside of them (use directives instead).
Use the controllers just to prepare the model. whenever possible delegate all the logic of your app to services. If you do so, it won´t really matter that much if your controller is view-centric or model-centric.
As I said before this is a very subjective matter and I´m sure many people will disagree with me.
I hope this could help you.
Basic Idea
So I'm actually in the process of migrating a legacy code base over to a restful based web service architecture utilizing AngularJs
Controllers are responsible for managing the data that is consumed by the view aka the webpage. My personal preference is that there is a one to one relationship between the controller and the view that it is serving. So, based on the question, Angular controllers should be more view centric. I'm sure that there are plenty of people who will disagree with me though.
If you are worried about the extensibility of this pattern, you should place your business logic and data access within Angular Services as described here. This offers you a tremendous amount of reuse of business logic operations as well as unit testability.
TLDR;
The specifications for Angular are changing all the time and with each new version there will be a new standard. A more view centric architecture looks appropriate for this application.
For some more complete reading on the subject I recommend checking out:
3 Tips To Building Enterprise Grade Angular/Node Applications
Lessons Learned: A Year with a Large AngularJS Project
I would highly appreciate it if someone can share their thoughts on this. I am new to Spring, but not new to MVC. However, my exposure with MVC has mostly been on the front-end side of things.
Lets say if I have a dynamic site which employs SpringMVC Framework 3.0, and site has 5 links such as:
HOME | FAMILY | COMEDY | HORROR | ACTION
Each of these links would query a database and display information. If I was to be doing this for front-end via a MVC framework, I would have five different controllers, but I came across a question here (Stackoverflow) which talks about having just one controller and then numerous resolvers. I would like to know what is the correct approach?
Here is the link to the mentioned question:
Spring controller setup question?
As Jaanas said, it really depends on how you are retrieving the data for each link and whether the resulting pages are all based on the same data model and templete (JSP) or are completely different to each other.
I tend to use a separate controller for "home" but if those other pages are simply querying a database then, I'd start serving those from a single controller.
Its also a good idea to try and think of your pages in terms of REST principles. Once you start seeing the links in terms of REST URIs, then the decision of whether the various URIs go in the same or separate controllers is obvious. E.g. if the URI for "family" is /movie/genre/family, then it becomes clear that all 4 of those links should go in the GenreController class (or a MovieController class, with a /genre method capturing the category with a #PathVariable annotation).
It depends on how much logic you have behind each link. If you have only one mapping in each controller, you could use single controller.
That said, I still prefer having separate controllers, when doing only small project like yours.
On the other hand, if your project expands, you could end up having absurd amount of controller, that don't do much. In that case you could group your controllers from start, so grouping similar controllers, that have almost equal logical side.
The number of controllers in your application depends on the scope of your project. If you have fixed requirement and the number of operations are limited then you can use single controller which will handle your all request.
But if your requirements are not fixed or going to increment with your application version, you must use separate controller for separate module. So that in future you can easily manage your application code.
For better code maintenance, I will prefer to use separate controller for each module.
I am contemplating about using an MVC pattern in my new project and I can clearly see the main advantage of being able to put the data layer (the model) a little closer to the presentation layer (the view), which will allow a little increase in application speed. But apart from performance stand point are there any other advantages of MVC over the view-logic-data layered type pattern?
EDIT:
For those who's interested I just uploaded a sample PHP code that I created to test the use of MVC. I purposly omitted all the security checks to make the code a little easier to read. Please don't critisize it too much, because I know it could be a lot more refined and advanced, but nevertheless - it works!!! I will welcome questions and suggestions: Here is the link: http://www.sourcecodester.com/sites/default/files/download/techexpert/test_mvc.zip
The separation of concerns that's quoted as being an advantage of MVC is actually also an advance of a 3-layer/3-tier system. There too, the business logic is independent and can be used from different presentation tiers.
A main difference is that in classic MVC the model can have a reference back to the view. This means when data is updated the model can push this data back to possibly multiple views. The prime example is a desktop application where data is visualized in multiple ways. This can be as simple as a table and graph. A change in the table (which is a change in one view) is first pushed via the controller to the model, which then pushes it back to the graph (the other view). The graph then updates itself.
Since desktop development is on the decline, a lot of programmers have only come in touch with MVC in some web variant, e.g. via JSF in Java EE.
In those cases the model almost never has a reference to the view. This is because the web is mainly request/response based and after a request has been served, the server cannot send additional information. I.e. an update pushed from the model to the client would be meaningless. With reverse ajax/comet this is changing, but many web based MVC frameworks still don't fully utilize this.
Thus, in the case of web based MVC, the typical "triangle" between M, V and C is less there and that MVC variant is actually closer to an n-tier model than 'true' MVC is.
Also note that some web MVC frameworks have an intermediate plumbing part between M, V and C called a backing bean (Java/JSF) or code behind (ASP.NET). In JSF the controller is provided by the framework, and the view often doesn't bind directly to the model but uses this backing bean as an intermediary. The backing bean is very slim and basically just pre-fetches data from the model one way and translates model specific messages (e.g. exceptions) into view specific messages (e.g. some human readable text).
Beside
code reuse,
separating of concerns,
less coupling between the layers,
already mentioned by #bakoyaro and #arjan
i think that MVC is better than 3-tier when combined with the "convention over configuration" pattern. (i.e. "ruby on rails" or Microsofts "MVC for asp.net").
In my opinion this combination leads to to better and easier code maintanance.
In the first place it makes learning the mvc-framework a bit more difficuilt since you have to learn the conventions (a la controllers go into the controllers folder and must be named xxxxxcontroller)
But after you learned the conventions it is easier to maintain your own and foreign code.
Forget increasing application speed by moving to MVC. I have found the biggest benefit to be ease of code reuse. Once you move to MVC, there are no dependencies on the presentation of your data or the storage of the actual data.
For example you could write a servlet that served up .jsp pages as your presentation layer one day, and the next day write a web service as another presentation layer to your existing Model and Controller. Like wise if you want or need to switch your DBMS. Since accessing the Model is completely separate from everything else, you would just need to re-write just your data access objects to return the data in a way your Controller can handle it.
By separating concerns into 3 distinct pieces, you also facilitate true unit testing. Your Presentation layer can be tested free of the Model or Controller, and vice-a-versa.
On a side note, I've often felt that the MVC abbreviation was inaccurate. Whenever I see it I think of it as View->Controller->Model. The presentation layer will never have DAO code in it, and the model will never have presentation logic in it. The Controller is forced to act as a go-between.
Where 3-tier separates presentation from business and data access, MVC is a presentation layer pattern which further separates Model (data) from View (screen) and Controller (input).
There is no choosing MVC over 3-tier/3-layered. Use them both.
Question number three in my quest to properly understand MVC before I implement it:
I have two cases in mind:
The primary application
window needs to launch the
preferences window. (One View
invoking another View.)
The primary Model for an application
needs to access a property
in the preferences Model. (One Model
accessing another Model.)
These questions are related in that they both involve communication across Model-View-Controller triplets, a topic that I haven't found much discussion of in my Googling.
The obvious way to fix this is to wrap everything in a top-level "application" object that handles transactions between Models and allows Controllers to invoke one another's methods. I have seen this implemented, but I'm not convinced its a good idea. I can also see possibilities involving Controllers observing more than one Model and responding to more than one View, but this seems like its going to become very cluttered and difficult to follow.
Suggestions on how best to implement this sort of cross-talk? I feel like its a very obvious question, but I've been unable to find a well-documented solution.
On a broader note, if anyone has a link that shows typical approaches to these sorts of MVC issues, I would love to see it. I haven't had much luck finding solid, non-trivial references. Examples in Python would be lovely, but I'll gladly read anything.
Edit 1:
I see some pretty interesting things being said below and in general no-one seems to have a problem with the approach I've described. It is already almost a lazy form of the FrontController design that Vincent is describing. I certainly don't foresee any problems in implementing that pattern, however, it doesn't seem that anyone has really addressed the question in regards to communication amongst Models. All the answers seem to be addressing communication among objects in a single Model. I'm more interested in maintaining separate Models for separate components of the application, so that I'm not stuffing fifty state properties into a single Model class. Should I be maintaining them as sub-Models instead?
With respect to (1), views don't invoke other views. They invoke controller actions that may result in other views being rendered. In your case, the primary application window contains a user interface element (button, link) that invokes a controller action to display the preferences window.
With respect to (3), model components certainly could be related to one another. This isn't unexpected nor to be avoided, necessarily. For instance your Customer model may have an associated set of Orders. It would be perfectly natural to access the customer's orders via a method in the Customer class.
You might want to take a look at the MVC page on wikipedia for an overview.
You may want to consider looking up the Front Controller design pattern.
The Front Controller pattern defines a single component that is responsible for processing application requests. A front controller centralizes functions such as view selection, security, and templating, and applies them consistently across all pages or views. Consequently, when the behavior of these functions need to change, only a small part of the application needs to be changed: the controller and its helper classes.
This way all requests from the view goes to the FrontController who then decides which specific action (controller) to invoke. Sometimes, it could forward straight to another view as in your first case.
There is no problem with multiple objects in the model talking to each other. In fact that will be very common. The way I see it, all the objects in the Model act like one component to represent the data and operations on the data.
This article might help. And this one.
Model does not mean a single model object. The model is the subset of the entirety of your domain model which is directly related to the controller actions and the views in question.
I just read a blog post that explains MVC with a banking analogy. I have a few months of experience with web application development with an MVC framework (CakePHP), so I get the basics, but I began to see a theme that made me think I'm taking a flawed approach to where I put my logic:
Fat models, skinny controllers
Keep as much business logic in the models as possible
In my app, models are anorexic and controllers are obese. I have all business logic in the controllers and nothing besides associations and validation rules in the models.
Scanning through my controllers, I can now identify a lot of logic that should probably go in a model:
The app has lists, which contain items, and the items can be ranked. The sorting logic which puts the list in ranked order is in a controller.
Similarly, items (Item model) also have images (Image model). Each item may have a default image (designated by image_id in the items table). When an item is displayed with its images, the default image should appear first. I have the logic that does this in a controller.
When a list is displayed, related lists are displayed in the sidebar. The logic to determine which lists are related is in a controller.
Now to my questions:
With the examples I gave above, am I on the right track in thinking that those are instances of logic presently in a controller that belongs in a model?
What are some other areas of logic, common to web apps, that should go into models?
I'm sure identifying this problem and changing my design pattern is half the battle, but even if I decide to take those examples I gave above and try to move that logic to a model, I wouldn't know where to begin. Can anyone point me in the right direction by posting some code here, or linking to some good learning resources? CakePHP specific help would be great, but I'm sure anything MVC will suffice.
It's a bit tough to give you the "right" answers, since some of them deal with the specifics of the framework (regardless of the ones you are working with).
At least in terms of CakePHP:
Yes
Anything that deals with data or data manipulation should be in a model. In terms of CakePHP what about a simple find() method? ... If there is a chance that it will do something "special" (i.e. recall a specific set of 'condition'), which you might need elsewhere, that's a good excuse to wrap inside a model's method.
Unfortunately there is never an easy answer, and refactoring of the code is a natural process. Sometimes you just wake up an go: "holy macaroni... that should be in the model!" (well maybe you don't do that, but I have :))
I'm using at least these two 'tests' to check if my logic is in the right place:
1) If I write a unittest, is is easy to only create the one 'real' object to do the test on (= the object that you are using in production) and not include lots of others, except for maybe some value objects. Needing both an actual model object and an actual controller object to do a test could be a signal you need to move functionality.
2) Ask myself the question: what if I added another way to use these classes, would I need to duplicate functionality in a way that is nearly copy-paste? ... That's also probably a good reason to move that functionality.
also interesting: http://www.martinfowler.com/bliki/AnemicDomainModel.html