I asked this question to get some opinions on the subject of glue code.
For example, imagine you have a class (pseudocode):
class MyClass
int attribute a
string attribute b
And to represent that data model, you have BOTH a slider and a text box to represent a, and a text box and say... the window label to represent b.
Obviously, when one of these view objects is changed, you want to update the others. However, updating the entire view is obviously inefficient.
method onSomethingHappened(uiObject)
model.appropriateAttribute = uiObject.value
The question is, what is your opinion on what to do next? Should the model object implement a callback that notifies a listener when the value has been changed, allowing one to write glue code like:
method modelChangedCallback(model, attribute)
uiObject1.value = model.a
uiObject2.value = model.a
Where you might examine what the attribute that changed is, and respond accordingly? This is the model in Objective-C and Cocoa on Mac, for the most part.
OR, would you rather have the responsibility lie completely in the glue code?
method onSomethingHappened(uiObject)
model.appropriateAttribute = uiObject.value
self.updateForAttribute("appropriateAttribute")
Both of these approaches can get pretty hairy (as is the problem with glue code) when your project gets large. Maybe there are other approaches. What do you think?
Thanks for any input!
For me I think it comes down to where the behavior is needed. In the situation you describe, the fact that you are binding multiple controls to a property is what is driving the requirement, so it doesn't make sense to add code to the model to support that.
In a web-based model I would probably put the logic in the web page since that can be done rather cheaply using Javascript. If I don't have that luxury (i.e. I'm dealing with a "dumb" view), then it would probably make sense to do it in the controller, or model glue code. If this sort of thing becomes common enough, I may go as far as creating some form of generic helper to reduce the amount of boiler-plate code I have to deal with.
Related
Given that there is file selection widget on the view and controller need to handle event of selecting file, should I rather write controller method:
public void fileSelected(String filePath){
//process filePath
}
or
public void fileSelected(){
String filePath = view.getSelectedFilePath();
//process filePath
}
The first approach seems to introduce less coupling between C and V: C don't know what exactly data does C need while handling given event.
But it requires creating a lot of verbose methods similar to getSelectedFile on V side.
On the other hand, second approach may lead to cluttered controller methods in more complex cases than in example (much more data to pass than just filePath).
From your own experience, which approach do you prefer?
The first approach is my favourite. The only difference is I would rather use an object (like Mario suggested) to pass arguments to the method. This way method's signature will not change when you add or remove some of the arguments. Less coupling is always good :)
One more thing:
If You want to try the second solution I recommend using a ViewFactory to remove view logic from the controller.
The first approach is the way to go;
public void fileSelected(String filePath){
//process filePath
}
The Controller should not care about how the View looks like or how it's implemented. It gets much clearer for the developer as well, when creating/updating the view, to know what an action in the controller wants. Also it makes it easier for method overloading.
Though, I don't know really how String filePath = view.getSelectedFilePath(); would work. Are we talking about parsing the View code/markup?
On the other hand, second approach may lead to cluttered controller methods in more complex cases than in example (much more data to pass than just filePath).
That's when you would create a View Model class (let's say we name it MyViewModel) to store all the properties that you need to send (may it be 10 properties) and then pass that in the action: fileSelected(MyViewModel model). That's how it's intended to be used and what the *ModelBinder's in asp.net mvc are there to help you with.
I think you need to look at this from a step back.
Worry less about how it gets in, and be more concerned with validation and error raising.
Tomorrow, your requirements could change and demand that you source the information via a different architectural approach. You could refactor the setup of [inputs / an input object] into a base controller class - or one of several classes for different controller domains.
If you focus on proper validation, whether within the controller (scrubbing) or outside of it (unit tests), then you perform more thorough decoupling though duck typing.
I would go with the first approach. It's reusable and separates concerns. Even if the method of getting the filePath in the future were to change, it won't affect your method's functionality.
Could anyone give an example of why it would be advantageous to use MVC instead of a simpler Model and a View only.
Note: whether it's called MVC or MVP (Model-View-Presenter), I'm talking about the one where the View receives input, then the Controller will respond to the input event by interpreting the input into some action to be done by the Model. When the model changes, the View will update itself by responding to events from the model.
What is disadvantageous of simply letting the Model respond to events in the View and vice versa?
In MVC, if I changed the model in a way that affects the controller then I'll have to do changes in the controller. In Model-View, if I change the Model, I'll have to update the view.
So, it seems like we are introducing complexity by adding the "controller" part?
In MVC, the Model is blind to its environment, the view can be too - passing off (blindly) its events to the controller, which knows more about the view and model. So when all is said and done, the controller is the 'non-reusable' disposable part of the system, since it is the most context aware component.
if I changed the model in a way that affects the controller...
The the model should expose simple CRUD methods in such a way that those using the methods do not have to know anything about the passed update object, nor what really happens inside the model.
This means that the view, IMO, has to do a bit of work by creating the passed record, since Controllers are supposed to be stateless and the view is more persistent. Controllers get triggered and 'kick-in' do their work with a passed object and do not have a state.
The passed data is created by some sort of generic convention.
Let me go even further. Suppose you have a view, a tablegrid, and a control whose enabled property is dependent on item is selected in the grid -- you COULD create a view that handles both those controls and this logic internally, and that would probably be the way to go in such a simplified example.
But the more atomic your views are, the more reusable they become, so you create a view for every, yes every, control. Now you are looking at a situation where views have to know about each other in order to register themselves for the right notification...
This is where the controller steps in, since we want to stick all these dependencies onto him, the long term disposable one. So the controller manages this type of view-to-view notification scheme.
Now your views are ignorant as they can be and independent, thus reusable.
You can code a view without having to know about the system, or the 'business logic' as they like to call it. You can code a model without having to know too much about your goals (though it does help to tweak the model to enable it to return the datasets you have in mind).... but controllers, they are last and you have to have the previous two firmed up before you can wire things together.
Here is another thing to think about -- just as the Model is supposed to abstract-away and provide a generic interface to the underlying implementation of the data it is managing (the client does not know if the data comes from a DB, a file, a program setting, etc) -- the view should also abstract away the control it is using.
So, ultimately this means a view should not (caveat below) have functions/properties that look like this:
public property BackgroundColor{get;set}
Nor
public function ScrollBy(x,y){}
But instead:
public SetProp(string name, object val){}
And
public DoCmd(string name, object val){}
This is a bit contrived, and remember I said ultimately... and you ask why is this a good idea?
With reusability in mind, consider that you may one day want to port things from WinForms to, say, Flex, or simple want to use a new-fangled control library that may not expose the same abilities.
I say 'port' here, but that is really not the goal, we are not concerned with porting THIS particular app, but having the underlying MVC elements generic enough to be carried across to a new flavor -- internally, leaving a consistent and ability-independent external interface intact.
If you didn't do this, then when your new flavor comes along, all your hard references to view properties in the (potentially reusable/refactorable/extendable) controllers have to be mucked with.
This is not to mean that such generic setters and cmds have to be the interface for all your views abilities, but rather they should handle 'edge case' properties as well as the normal props/cmds you can expose in the traditional hard-link way. Think of it as an 'extended properties' handler.
That way, (contrived again), suppose you are building on a framework where your buttons no longer have buttonIcon property. Thats cool because you had the foresight to create a button view interface where buttonIcon is an extended property, and inside the view your conditional code does a no-op now when it receives the set/get.
In summary, I am trying to say that the coding goals of MVC should be to give the Model and View generic interfaces to their underlying components, so when you are coding a Controller you don't have to think to hard about who you are controlling. And while the Controllers are being (seemingly unfairly) set up to be the sacrificial lamb in the long run of re-usability -- this does not mean ALL your controllers are destined for death.
They are hopefully small, since a lot of their 'thinking' has been shoved off into semi-intelligent Models and Views and other controllers (ex: Controller to Sort a Grid or Manipulate a TreeView) -- so being small they can be easily looked at and qualified for reuse in your next project -- or cloned and tweaked to become suitable.
It actually reduces complexity by separating the workflow logic from the domain logic. It also makes it easier to write unit tests and makes your application easier to maintain and extend.
Imagine if you wanted to add a new data type. With the approach above, you would probably duplicate a lot of the workflow logic in the new class as it would be likely to be tightly coupled to the domain logic.
The discipline involved in separating the workflow logic into the controller makes it more likely that you will have fewer dependencies between workflow and domain logic. Adding a new data type would then be more simple, you create the new domain object and see how much of the controller you can reuse, e.g. by inherited from a controller super class.
It would also make it easier to change frameworks in future - the model would probably not change too much and so would be more portable.
Having said that, you might want to look into MVVM depending on what you are using as your presentation layer: Benefits of MVVM over MVC
Advantages of MVC/P (I am talking about Supervising Controller here) over MV include:
You can handle complex data binding code in the controller, if required.
You can test that complex presentation logic without a UI testing framework.
You can also have a graphic designer make your views, and not see your code, and not mess up your code when they fix your views.
I have another programmer who I'm trying to explain why it is that a UI component should not also be a data-structure.
For instance say that you get a data-structure that contains a record-set from the "database", and you wish to display that record-set in a UI component within your application.
According to this programmer (who will remain nameless, he's young and I'm teaching him...), we should subclass the data-structure into a class that will draw the UI component within our application!!!!!!
And thus according to this logic, the record-set should manage the drawing of the UI.
******Head Desk*****
I know that asking a record-set to draw itself is wrong, because, if you wish to render the same data-structure on more than one type of component on your UI, you are going to have a real mess on your hands; you'll need to extend yet another class for each and every UI component that you render from the base-class of your record-set;
I am well aware of the "cleanliness" of the of the MVC pattern (and by that what I really mean is you don't confuse your data (the Model) with your UI (the view) or the actions that take place on the data (the Controller more or less...okay not really the API should really handle that...and the Controller should just make as few calls to it as it can, telling it which view to render)) But it's certainly alot cleaner than using data-structures to render UI components!
Is there any other advice I could send his way other than the example above? I understand that when you first learn OOP you go through "a stage" where you where just want to extend everything.
Followed by a stage when you think that Design Patterns are the solution every single problem...which isn't entirely correct either...thanks Jeff.
Is there a way that I can gently nudge this kid in the right direction? Do you have any more examples that might help explain my point to him?
Have you heard of Martin Fowler?
Separating User Interface
Code
Anyway, if he wants to go further in that direction of adding render methods to his data controls, have him look at "loose coupling". It's okay to create some generic type of interface that gets him halfway there, but the UI component should take it the rest of the way.
This boils down to functional vs. non-functional responsibilities. What the data structure does and how it's visualized are two completely separate things -- essentially the root of the MVC pattern.
There's also a notion of circular dependencies here. Since the UI must know about the data structures, if you allow the data structures to then depend on the UI, you've got yourself a nice little ball of mud.
Generally on the point of decoupling:
Not only can there be different components of the UI rendering the same data structure. You may even have completely different UIs (Web, Desktop Application, ...) Now of course, you could subclass Person with WebPerson and DesktopPerson (this already sounds wrong, doesn't it? The naming is simply not about the kind of Person - it's about something else).
Each UI could work on different kinds of Persons, e.g. Teacher and Student. So we get WebPerson, WebTeacher, WebStudent, DesktopPerson, DesktopTeacher and DesktopStudent.
Now let's say, WebPerson defines the method "drawAddressFields()" to draw a web version of the address fields. But since WebTeacher has to derive from Teacher to use the additional data field "salary" (and let's assume single inheritance), it must implement "drawAddressFields()" once again!
So maybe the argument of "this will cause much more work" will help to create some motivation :-)
BTW, it will automatically lead to creating some delegate that implements the code of drawAddressField(), which will then evolve to creating a component that does the drawing separately from the data structure.
After writing a few lesser programs when learning Java the way I've designed the programs is with Model-View-Control. With using MVC I have a plethora of getter methods in the model for the view to use.
It feels that while I gain on using MVC, for every new value added I have to add two new methods in the model which quickly get all cluttered with getter & setters.
So I was thinking, maybe I should use the notifyObserver method that takes an argument. But wouldn't feel very smart to send every value by itself either so I figured, maybe if I send a kind of container with all the values, preferably only those that actually changed.
What this would accomplish would be that instead of having a whole lot of getter methods I could just have one method in the model which put all relevant values in the container.
Then in the view I would have a method called from the update which extracted the values from the container and assigning them to the correct fields.
I have two questions concerning this.
First: is this actually a viable way to do this. Would you recommend me doing something along these lines?
Secondly: if I do use this plan and I don't want to keep sending fields that didn't actually change. How would I handle that without having to have if statements to check if the value is not null for every single value?
I've more familiar with the MVP paradigm, but hopefully they're similar enough to comment. While getters (and setters) in and of themselves are not necessarily evil, they are sometimes a sign that your subsystems are too strongly coupled. One really great way to decouple this is to use an event bus: see Best practices for architecting GWT apps. This allows the view to just shoot off events for the controller to listen for whenever something important happens, and the view can listen for events whenever something changes in the model that corresponds to updating the view. Ideally you wouldn't even need to ever pass the model to the view, if you can break up any changes into incremental pieces and just tell the view to change this part and then this other part.
If you feel you have too many getters (and setters) in your model class, maybe you have too many fields altogether. Is it possible that there are several distinct classes hiding within your model? If you extract these into separate classes, it may make your model more manageable.
OTOH the associated container you are thinking about could also be viable - but then why duplicate all data? You could instead use the associated container directly in the model to store all properties you can think of. And you can also pass this around for observers to get updates (preferably wrapped into an unmodifiable container, of course) - although in this setup you wouldn't need to.
In general, Java is a verbose language which expects you to put all those getters and setters (and a lot more) in place. However, any decent IDE can generate those for you with a few keypresses. Note also that you need to write them only once, and you will read and call them many many more times. Verbose also means easily readable.
If you have too many getter it's ok. But you shouldn't need the setter. The view is supposed to only read/query the model.
The MVC pattern should promote something that is asymmetric: the control update the model by calling methods in the model that embed the logic and update the sate accordingly; this respects encapsulation. The view reads/queries the model via the getters. This goes a bit against information hiding, but that's how MVC works.
I wouldn't personally pass all information in the events. It sounds complicated to me: either you end up with something that is not statically typed (e.g. you pass hashmaps), or with a plethora of typed events. I would stick with something simple, and have (possibly many) getter in the model.
I am writing a GUI based application where I read a string of values from serial port every few seconds and I need to display most of the values in some type graphical indicator(I was thinking of QprogressBar maybe) that displays the range and the value. Some of the other data that I am parsing from the string are the date and fault codes. Also, the data is hierarchical.
I wanted to use the model/view architecture of Qt because I have been interested in MVC stuff for a while but have never quite wrapped my brain around how to implement it very well.
As of now, I have subclassed QAbstractItemModel and in the model I read the serial port and wrap the items parsed from the string in a Tree data structure. I can view all of the data in a QtreeView with no issues.
I have also began to subclass QAbstractItemView to build my custom view with all of the Graphical Indicators and such. This is where I am getting stuck. It seems to me that in order for me to design a view that knows how to display my custom model the view needs to know exactly how all of the data in the model is organized. Doesn't that defeat the purpose of Model/View? The QTreeView I tested the model with is basically just displaying the model as it is setup in the Tree structure but I don't want to do that because the data is not all of the same type. Is the type of data or the way you would like to present it to the user a determining factor in whether or not you should use this architecture? I always assumed it was just always better to design in an MVC style.
It seems to me like it might have been better to just subclass QWidget and then read in from the serial port and update all of subwidgets(graphical indicators, labels, etc...) from the subclass. Essentially, do everything in one class.
Does anybody understand this issue that can explain to me either what I am missing or why I shouldn't be doing it this way. As of now I am a little confused.
Thanks so much for any help!
** ** - Is this a situation where Qt Model/View architecture is not useful?
I"m going to say not necessarily - end edit
I'm not sure I fully understand your question, but let me try.
First, let's talk about MVC display pattern. This pattern is all about breaking the program into separate (and hopefully testable) sections that have their own areas of concerns.
The model is data structures that describe your data. I'm sure that there is some statistical data is also present. It is important that the model not know anything about how the data is going to be displayed to the user.
The View is how information is presented to user. It is not suppose to care about how the data is presented for displaying. It is important for this layer to be unaware of how the model gets the data for display.
The control logic is the "glue" to connects the first two items together. This is the layer that holds all the "messy" stuff to make to good user experience. EDIT - Most of what QT calls a "view item" I would probably put in the controller layer.
But if you do this, then the model and controller layers become very testable.
With that being said, many of your points are not really related to MVC pattern. You seem to be discussing what is the optimum way to display the data. This is always a problem. And without seeing your app, I'm not really going to try to tell you what is going to look good.
But by following good MVC pattern design, you can make pretty significant revision to the display without effecting the underling code.
This being said, I'm dealing with this exact issue right now and this pattern is working well form me. If you go to codeplex.com and search for mvvm (model-view-viewmodel, term used in WPF), you will see a number of projects that use it that you can use to get more information.
If this is not enough, let me know and I may be able to give you a better answer.