I have a class that I would like to extend by defining a new class that contains the first class as a public property, as well as additional added properties. However, the class that I'm extending has multiple derived types, which should be treated the same in the extension class.
Below is an example of what I am trying to do:
Public Class ClassA
End Class
Public Class ClassB
Inherits ClassA
End Class
Public Class ClassC
Inherits ClassA
End Class
Public Class BaseExtended
Public Property Foo As ClassA
Public Property ExtendedMetaData1 As Double
Public Property ExtendedMetaData12 As Integer
End Class
Public Class DerivedExtendedB
Inherits BaseExtended
Public Property Foo As ClassB
End Class
Public Class DerivedExtendedC
Inherits BaseExtended
Public Property Foo As ClassC
End Class
The code that uses an instance of any of the 'extended' classes would then need use that instance appropriately depending on it's type. There would be many cases where the property 'Foo' needs to be accessed and modified outside of the class that it belongs to.
If I were to implement something like what I have shown above, that would require that I first cast it to the required type before accessing or modifying it. Ideally I would like to do that inside the 'DerivedExtended' class; The alternative, I think, would be to duplicate code to cast that property would [hundreds of times] in the client code.
Private Sub ClientUsesObject(bar As BaseExtended)
' Perform a task that is agnostic Foo type
' Would not require that Foo be cast to any specific type
If bar.GetType() Is GetType(DerivedExtendedB) Then
Dim barCast As DerivedExtendedB = DirectCast(bar, DerivedExtendedB)
' Perform task that requires Foo to be of type ClassB
ElseIf bar.GetType() Is GetType(DerivedExtendedC) Then
Dim barCast As DerivedExtendedC = DirectCast(bar, DerivedExtendedC)
' Perform task that requires Foo to be of type ClassC
End If
End Sub
What I'm looking for is advice outlining or describing a design pattern that can handle this situation. I've searched for quite a while, and have not been able to find any examples that solve this problem.
I realize that this may be somewhat of an "XY" problem. I'm working with existing code that simply assumes all instances are of the same derived type (when in fact some instances are of the other derived type). As such, the existing code does not work. To me what I've tried to outline above seems like the most straightforward path, but I'm open to alternative if this is just the wrong approach.
This pattern of type covariance in derived classes is the canonical reason for what is called in C++ the "Curiously Recurring Template Pattern" and has been called in .NET the "Curiously Recurring Generic Pattern." I believe it's also sometimes referred to as "F-Bounded Polymorphism" (not a computer scientist, so I might have the reference wrong).
You can write a base class like this:
Public Class Base(Of TDerived As Base)
Public Overridable Property foo As TDerived
End Class
And then use it like this:
Public Class MyDerived
Inherits Base(Of MyDerived)
End Class
Then, the derived class has a property foo whose type is MyDerived. No casting required by clients.
However, this has some limitations. It works best when you don't need to switch back and forth between derived and base. There is no one Base, so you can't declare instances of it. If you want to be able to declare something as Base, then you end up needing to fall back on a non-generic base class. This will still work well for certain usage patterns where you don't need to convert from base to derived, but otherwise you run right back into the casting problems you are trying to avoid.
Eric Lippert has written a bit about this pattern. He's always interesting to read, so I'd recommend looking up his commentary.
Another alternative to consider, if the generic approach doesn't work for you, is code generation. You can use T4 templates to process a compact description of what your code should be, and generate the code files from them. A long list of casts is less tedious if you only write the machinery to generate it, you don't write them all out explicitly.
Related
Assume we have a class called "MyClass"
Public Class MyClass
End Class
this class has a function called "My function"
Public Class MyClass
Public Function MyFunction()
End Function
End Class
This class has been implemented for some time and its been working fine. Now we need to change the implementation of the function "MyFunction". One option would be to open the source code and change it there. But I'm guessing there has to be a better approach.
Inheritance comes to mind but I don't want to change the derived classes name. I want the name of the class to still remain "MyClass", But I'm guessing the code below will cause an error:
Public Class MyClass
Inherits MyClass
Public Function MyFunction()
End Function
End Class
In other words I'm trying to create a new version of the old class by keeping most of the members the same but just changing a few functions.
To explain the project as a whole, The program is meant for structural design. What it does it designs structural components (i.e columns, beams, slabs, ...). The design procedures are specified by 3rd parties (government regulations). For example:
In the year 2007 government regulations specified that column dimensions are to satisfy the equation F:
H*B < Fy^2/L
In the year 2008 they introduced a new function G and they say column dimensions must satisfy this new function:
H*B^2 < Fy^0.5/E+Alpha^2/L
Where H and B are column dimensions.
What I don't want to do is to open the source code every year and make these changes. I want to somehow override the functions that need to be changed without opening the source.
Any Ideas?
The code is generally not supposed to be changed over time. That is - if you wrote code that is guaranteed to break after 2 weeks by itself, you probably should reconsider your design.
As you rules/regulations come out, you usually update your input data (in a form of XML, or a relational database for large amounts of data), and your program would automatically pick those up.
The only case you would update your program under this scenario is when new type of regulations come out. But even in this case the changes are usually minimal.
A good anti-pattern example for this - you have 500 forms, each of them has 500 lines of code, so that's 250000 lines of code in your UI layer. New regulations come out that requires changing 50% of the code in each form. Your impact is 125000, which at 40 lines of code per day would take 8.5 developer-years.
A solution to this would be having a change of 100 lines spread across all forms, adding 1 line in each, or leaving everything as is. Also there will be a data load/conversion procedure from a government/other file, which populates your database in the proper format, updating the values or adding new ones. There may be 10 lines of change in that program, but that's about it, 3 days worth of work, if you believe in 40 LoC per day. Otherwise it still falls under 2 weeks of developer's time.
Depending on how you implement it, the benefit of this approach could be that you support old standards as well, so older input can be matched and production reports can be generated. It is a good practice to be able to back-date your reports, cause sometimes there are issues in report code left unnoticed for months before being discovered.
EDIT: A more structured approach to what I suggested in the comments would be storing expression trees in the DB. Most simple form of it is just a linear workflow, using postfix notation (single table). For example A, B, + C - is equivalent to A + B - C. You can then have a user interface for some configuration tool, which only allows user to input values and functions that are applicable. This is assuming applicable values are also stored in DB as parameters (one structural component can have 0...N of them).
Inheritance can do what you want but you need to create a new ancestor, not descendant.
Change the name of the original class to something that denotes that it is a base class. Also, add the MustInherit modifier to the class and Overridable to any of the methods or properties that you may need to override.
One thing to watch for is Private members in this base class. Any members that need to be accessible from the descendant class cannot be Private and must be changed to Protected.
The original class looks like this.
Public MustInherit Class MyBaseClass
Public Overridable Function MyFunction() As String
' code...
End Function
Public Overridable Function AnotherFunction() As String
' code...
End Function
End Class
Now create a new class with the original class's name which inherits from the base class. Override just the members that need to be different.
Public Class MyClass
Inherits MyBaseClass
Public Overrides Function MyFunction() As String
' new code...
End Function
End Class
That will get you started. The Template Pattern will allow you to do more fine grained code changes where only parts of a method need to be changed.
The formula is a bit complicated and you'll still need to change some code unless you store these in a database somehow.
An option would be to use inheritance with a factory method.
Public Class BaseClass
Public MustOverride Function MyFunction()
Public Function GetInstance(ByVal year As Integer) As BaseClass
If year = 2007 Then Return New Class2007()
If Year = 2008 Then Return New Class2008()
End Function
End Class
Public Class Class2007
Inherits BaseClass
Public Overrides Function MyFunction()
' H*B < Fy^2/L
End Function
End Class
Public Class Class2008
Inherits BaseClass
Public Overrides Function MyFunction()
' H*B^2 < Fy^0.5/E+Alpha^2/L
End Function
End Class
then, everywhere in your code you use BaseClass never knowing that Class2007 and Class2008 exists
Dim o As BaseClass
o = BaseClass.GetInstance(2007)
o.MyFunction()
Depending on the need, this can also be done with interface.
If you need to store the formulas in the database as string, you'll need to get a parser and this can also be found using 3rd party library. https://stackoverflow.com/questions/1387430/recommended-math-library-for-c-net
I am getting used to using interfaces, generics and develping them using inheritance in a real envrionment whilst trying use and implement this into a new architecture for one of our upcoming projects and I have a question regarding generics which I am confused about.
This is more of a educational question for myself because I can't understand why .NET doesn't allow this.
If I have a generic class which is (Of T As IA, T2 As A) then I have the following interfaces and class which implements the base interface
Public Interface IA
Property A As String
End Interface
Public Interface IB
inherits IA
Property B As String
End Interface
Public Class GenericClass(Of T As IA, T2 As A)
'Should be list of IA?
Public list As New List(Of T)
Public Sub Add()
End Sub
End Class
Because I have made T as IA why in the add method is Dim foo4 As T = New A() not legal when
Dim foo1 As IA = New A()
Dim foo2 As T
Dim foo3 = Activator.CreateInstance(Of T2)()
Dim x As IA = foo2
Dim y As IA = foo3
list.Add(x)
list.Add(y)
All of the above is? This is becoming a learning curve for me with generics etc. but I am just very confused with why I logically can't do this?
EDIT: Sorry forgot Class A and error message please see below
Public Class A
Implements IA
Public Property A As String Implements IA.A
End Class
EDIT 2: Error was typed incorrectly
"Value of type class a cannnot be converted to T"
It's not exactly clear what you're trying to do, but one problem I notice is that you seem to be assuming that a List<TypeThatImplementsIA> is somehow interchangeable with a List<IA>. That is not the case. Imagine that A were a class of flying birds, and IA were implemented by creatures that can fly, and someone created a GenericClass<Airplane, BlueJay). Even though Airplane and BlueJay are both things that can fly, one would not be able to add a BlueJay to a List<Airplane>. The one common situation in the Framework where one can use a GenericType<DerivedType> as a GenericType<BaseType> is with IEnumerable<T>. The reason for that is that one can't store T's into an IEnumerable<T>--one can only read them out. If one is expecting an IEnumerable<Animal> and one is given an IEnumerable<MaineCoonCat>, then every time one expects to read an Animal, one will read an instance of MainCoonCat, which inherits from Animal and may thus substitute for it. This feature of IEnumerable<T> is called covariance.
There's a limitation to such behavior, though, which stems from the fact that there is a difference between using an interface as a type of storage location (variable, parameter, etc.), versus using it as a constraint. For every non-nullable value type, there are actually two related types within the Runtime. One of them is a real value type, which has no concept of inheritance (but can implement interfaces). The other is a heap-object type which derives from ValueType (which in turn derives from Object). Most .net languages will implicitly convert the former type to the latter, and allow code to explicitly convert the latter to the former. Interface-type storage locations can only hold references to heap objects. This is significant because it means that while a struct which implements an interface is convertible to that interface type, that doesn't mean instance of the struct is an instance of that interface type. Covariance works on the premise that every object returned by e.g. an IEnumerable<DerivedType> may be used directly as an instance of BaseType without conversion. Such direct substitutability works with inherited class types, and with interfaces that are implemented by class types. It does not work with interfaces implemented by struct types, or with generics that do not have a class constraint. Adding a class constraint to a generic class type parameter will allow that type parameter to participate in covariance, but may preclude the use of structs as the generic type parameter. Note that unless one has particular reason to expect that an interface will be implemented by structures (as is the case with e.g. IComparable<T>, in many cases it's unlikely that an interface would be implemented by a structure and thus a classconstraint would be harmless).
That's because T is not the interface IA itself. It is one implementation of it.
Suppose that you have another class that implements IA:
Public Class B
Implements IA
Public Property B_A As String Implements IA.A
Public Property OtherProperty as Object
End Class
Then you create a new instance of Generic Class like this:
Dim genericObject as new GenericClass(Of B, A)
So in this case, T now is B, and A cannot be casted to B.
In this case instead, replacing the part of your doubt, a code that would make sense for me:
Dim foo4 As IA = New T()
EDIT due to comment
To be able to instantiate T, it is necessary to declare the New constraint in the type definition. So the generic class declaration would be:
Public Class GenericClass(Of T As {New, IA}, T2 As A)
I'm working with vb.net, wcf, wpf and I'm refactoring working code with the hope of being able to reduce some amount of redundancy. I have a bunch of methods that get called in several places throughout the code that only have a slight variation from each other and I would like to replace them with a single method instead.
Specifically, each of the redundant methods process an 1-d array that contain different objects I have created. There are several of these different object types each with different signatures but they have all have a "name" and "Id" property. (Also these objects don't have a shared base class but I could add that if needed.) Each of the redundant methods deal with a different one of the object types.
To refactor the code I would like to pass any of the different object arrays to a single new method that could access the "name" and "id" properties. I'm trying to write this new method in a fashion that wouldn't require me to update it if I created more objects down the road.
I've done some reading on Delegates and Generic Classes but I can't really figure out how this fits in. It would almost be as if I wanted to create a generic class that could handle each of my object types but then somehow also access the "name" and "id" propeties of the different object types.
Any help you can provide would be appretiated. Also, please keep in mind this project is written in VB.net.
Thanks
Mike
It sounds like having your object implement a common interface or have a shared base class would be best. Interfaces give you the most flexibility down the road if you ever need to pass a class to this method that must derive from some other class that does not implement the interface. However, a base class that implements the interface may also be useful just to reduce the duplicate declarations of these properties.
Public Interface IThingThatHasNameAndId 'good name not included
ReadOnly Property Name As String
ReadOnly Property Id As Integer
End Interface
Once you have the interface, you can then pass arrays of types implementing the interface as IEnumerable(Of IThingThatHasNameAndId) or make a generic method taking T() and constrain T to the interface.
Make a base class with the Name and ID properties, then you can make a method that takes in any class that derrives from that class.
Public Function TestFunction(Of t As YourBaseClass)(Byval obj As t) As Boolean
If obj.Name = "Some Name" AndAlso obj.ID = 1 Then
Return True
Else
Return False
End If
End Function
This is kind of two questions (one more specific than the other).
If I have a method like this:
Public Function Blah(String Foo)
End Function
Can I qualify Foo against another type (for instance can I require that Foo be a String that also implements IInterface?).
I'm imagining something vaguely similar to this:
Public Function Blah(RandomObject Foo Where RandomObject Is IInterface)
End Function
Additionally, is there any way to qualify the Type parameter?
For instance, can I require that the Type I take as a parameter is of a particular class tree?
Public Function Blah(Type t Where Type Of String)
End Function
I should mention that I am using this in the context of a property of an attribute so the class declaration itself cannot be generic (this is purely focused on qualifying a method parameter rather than typing a class and its methods).
This looks like a case for generics to me. Your method signature would be something like this in VB.NET:
Public Function Blah(Of T As {IImplementedByT})(Foo As T)
This specifies that Foo can be of any type T as long as T implements IImplementedByT. Note that this method can be generic without the containing class needing to be generic. If you want T to be a class derived from RandomClass that also implements this interface, you can specify both constraints:
Public Function Blah(Of T As {RandomClass, IImplementedByT})(Foo As T)
You can do the first for a generic type, but not for a nongeneric type. Basically a variable (including a parameter) can only have one compile-time type, so you can't say "it has to be a Foo and an IBar - you have to pick one or the other. Generics let you say "it has to be some type T where T derives from Foo and implements IBar" though.
Generics is a huge topic - too big to cover in a Stack Overflow answer - but Microsoft has a good introductory article.
As for your second question - no, you can't do that. The Type value will only be known at execution time, so it has to be an execution time check. You can write that check fairly easily though, with Type.IsAssignableFrom.
Not sure what you mean by "Foo be a String that also implements IInterface".
string class is sealed, so you can't inherit from it & hence you cant implement an interface on top of it.
I hope I am on the right page.
In VB.NET, I am trying to talk to a webservice (that can't be changed) to create and update customer data.
The CreateCustomer service expects an object of type ConsumerPerson and the ChangeCustomer service expects an object of type ChangeData.
The properties of these two object are exactly the same, so I thought it would be wise to just set the properties using one single function.
However, I am unable to find a way to tell my function that I want to fill either the ConsumerPerson or the ChangeCustomer object.
How do I make this work without late binding problems?
An interface sounds like your best approach. Here is a short code snippet. I picked a simple property named "Name" of type string. It should be easy to modify with the actual properties on your class.
Public Interface ICustomerData
ReadOnly Property Name As String
End Interface
Public Class ConsumerPerson
Implements ICustomerData
Public ReadOnly Property Name As String Implements ICustomerData.Name
Get
return _name
End Get
End Property
End Class
Public Class ChangeData
Implements ICustomerData
Public ReadOnly Property Name As String Implements ICustomerData.Name
Get
return _name
End Get
End Property
End Class
use an interface !
declare an interface IFoo, and implement its members in your subclasses ConsumerPerson and ChangeCustomer. That's exactly what interfaces are for.
You create an interface which both classes implements.
Is it not possible to overload your function with the second data type?
If you cannot change your objects, but they share the same field names, you could xml serialize the data and deserialize as the other class. - You should strongly consider the performance implications of this; however, it would give you the functionality you're asking for.