Please read 2nd Edition first.
Let's suppose that we have two dependent generic interfaces:
interface ITemplate1<T1, T2>
where T1 : ITemplate1<T1, T2>
where T2 : ITemplate2<T1, T2>
{
T2 t2 { get; set; }
}
interface ITemplate2<T1, T2>
where T1 : ITemplate1<T1, T2>
where T2 : ITemplate2<T1, T2>
{
T1 t1 { get; set; }
}
And from both of them two different classes have been implemented:
class Class1_1 : ITemplate1<Class1_1, Class2_1>
{
public Class2_1 t2
{
get;
set;
}
}
class Class1_2 : ITemplate1<Class1_2, Class2_2>
{
public Class2_2 t2
{
get;
set;
}
}
class Class2_1 : ITemplate2<Class1_1, Class2_1>
{
public Class1_1 t1
{
get;
set;
}
}
开发者_JAVA百科
class Class2_2 : ITemplate2<Class1_2, Class2_2>
{
public Class1_2 t1
{
get;
set;
}
}
Now, I want to define a class which implement ITemplate1
regardless of its required types..
class MyClass<T> where T : ITemplate1< ? , ? >
{
...
}
..to prevent from declaring more than one class as:
class MyClass1<T> where T : ITemplate1<Class1_1, Class2_1>
{
...
}
class MyClass2<T> where T : ITemplate1<Class1_2, Class2_2>
{
...
}
How can I achieve that? (i.e. Real problem may involved in more than two classes!) Thanks in advance.
1st Edition:
Class1_1
& Class2_1
as well as Class1_2
& Class2_2
must be used together. In other words, I'm looking for a way to have:
class MyClass<T>
where T : ITemplate1<Class1_1, Class2_1>
or
where T : ITemplate1<Class1_2, Class2_2>
{
...
}
2nd Edition: I think I could ask this question simpler to avoid to put dear readers into inconvenience. So I do apologize if I mixed you up. The reason for asking such a confusing question is that the real problem is completely complicated. Any way, here is the simple form:
Title: How to control implementation of interfaces?
Descriptions: I have a class which implements two dependent interfaces. I need to control given types, because only some pair of types are consistent with each other. --> 1st Edition
I don't know excactly if you are searching for exactly this, but you can make your ClassX_y more generic for example like:
class Class1_x<T2> : ITemplate1<Class1_x<T2>, T2>
where T2 : ITemplate2<Class1_x<T2>, T2>
{
public T2 t2
{
get; set;
}
}
class Class2_x<T1> : ITemplate2<T1, Class2_x<T1>>
where T1 : ITemplate1<T1, Class2_x<T1>>
{
public T1 t1
{
get;
set;
}
}
Or something like this:
class Class1_xy<T1,T2> : ITemplate1<T1,T2>
where T2 : ITemplate2<T1,T2>
where T1 : ITemplate1<T1,T2>
{
public T2 t2
{
get; set; }
}
class Class2_xy<T1, T2> : ITemplate2<T1, T2>
where T2 : ITemplate2<T1, T2>
where T1 : ITemplate1<T1, T2>
{
public T1 t1
{ get; set; }
}
class ClassBoth_xy<T1, T2> : ITemplate1<T1,T2>, ITemplate2<T1, T2>
where T2 : ITemplate2<T1, T2>
where T1 : ITemplate1<T1, T2>
{
public T1 t1
{ get; set; }
public T2 t2
{ get; set; }
}
But I don't see exactly the point in this - maybe you can enlighten us?
Using ClassBoth_xy<T1, T2>
is a really good idea but can not restrict MyClass
from being implemented by incorrect pairs of <T1, T2>
. I solved this issue as:
class MyClass<T1, T2>
where T1 : ITemplate1<T1, T2>, new()
where T2 : ITemplate2<T1, T2>, new()
{
public MyClass()
{
T1 _t1 = new T1();
T2 _t2 = new T2();
bool isValid = (_t1 is Class1_1 && _t2 is Class2_1) ||
(_t1 is Class1_2 && _t2 is Class2_2);
if( !isValid )
throw new Exception("Inconsistent types have been used to instantiate MyClass.");
}
...
}
精彩评论