I have the following two methods that both create me a couple of Constraint<T>
instances. The first creates types of Constraint<Quantity>
and the second creates types of Constraint<IParameter<Quantity>>
:
public void ConstrainBetween(Quantity min, Quantity max)
{
..
}
public void ConstrainBetween(IParameter<Quantity> min, IParameter<Quantity> max)
{
..
}
The argument types Quantity
and IParameter<T>
are like this:
public class Quantity
{
double Value;
..
}
public class IParameter<T>
{
T Value;
Func<T> Formula;
..
}
It all has worked nicely for months until i overloaded the implicit operator for IParameter as follows :
public class IParameter<T>
{
public T Value;
public Func<T> Formula;
public static implicit operator T(IParameter<T> p)
{
return p.Value;
}
}
Now the following usage no longer compiles:
HydrModel.TopWaterLevel5yr.ConstrainBetween(OutletPipe.IL, Transform.Z);
HydrModel.TopWaterLevel30yr.ConstrainBetween(OutletPipe.IL, Transform.Z);
Error:
The call is ambiguous between the following methods or properties: 'QuantityInput.ConstrainBetween(Quantity, Quantity)' and 'QuantityInput.ConstrainBetween(IParameter, IParameter)'
I understand that the compiler implicitly converts IParameter<Quantity>
to Quantity
making both method signatures potential targets and I know that I can resolve the ambiguity problem by:
- not overloading the implicit operator
- using an explicit operator overload instead
- using two different method names
I don't want to to 1 or 2 because the benefits elsewhere of implicit overloading outweigh this drawback.
I am currently intending to take strategy (3) and work with different signature names.
I post this question just in case there is an alternative approach that would allow me to have my cake and eat it? For example is there some way to specify exactly the method signature in the call to invoke it?
EDIT
Someone asked why do i need both methods - why not just get rid of the second one since the implicit overload will ensure calling of the first?
The reason is that calling Constraint<Quantity>.Value
always returns the same Quantity whereas calling Constraint<IParameter<Quantity>>.Value
returns a different Quantity at different times according to its Formula property. Therefore I do need a strategy for creating a Constraint where T can be either Quantity
or IParameter<Quantity>
.
If your
IParameter<Quantity>
is properly designed to be implicitly cast toQuantity
, then you shouldn't have any need in the second method ConstrainBetween. TheIParameter
also seems to be just a simple DTO, carrying a single value, and not anything else.So, my solution would be to get rid of
public void ConstrainBetween(IParameter<Quantity> min, IParameter<Quantity> max)
entirely and focus on the value in the first method, rather than on the value's carrying type.Update
Since @MonkeyFace has edited his post, I would also like to add some thoughts.
It seems that now
IParameter<T>
doesn't carry a single value anymore, but two or more values. The difference described in the updated question between the two methods also indicates thatIPerameter<T>
is not an equivalent to itsT
value.From this perspective it now seems odd to have an
implicit
operator, because (a) it loses that additional info (since losing information should only be done so deliberately - in my opinion) and (b) the two types are not equivalent to each other.So the new proposed change would be to migrate from implicit operattor to explicit one.
I see your comment that "there are drawbacks of using explicit operatior", but I imagine those drawbacks being having to use
(T)
in those scenarios where the cast should occur, and I consider those cases the lesser evil.