Question

In developing an F# application, I have a type that comprises a property of type Lazy<'T>.

Apparently, one interesting side effect (pardon the pun) of the way that F# handles the syntactical sugar of properties (as opposed to the C# way) is that the getter and setter of a property may return/accept different types. (At least, Visual Studio is not complaining as I write code that takes advantage of this observation.)

For example, it is advantageous for me to do this:

let lazyValue = lazy 0

member this.Value
    with get () =
        lazyValue.Value
    and set _lazyVal =
        lazyValue <- _lazyVal

… such that Value returns an int, but accepts only a Lazy<int>.

What I am wondering about are the theoretical, idiomatic, and practical objections to doing things this way. Is this something at which an F# snob would turn up his nose? Is there some functional programming rule of thumb that this (object-oriented implementation) clearly violates? Is this an approach that has been demonstrated to cause problems in large-scale applications? If so, why/how?

Answer 1

Perhaps it is a bug that “Visual Studio is not complaining as [you] write code that takes advantage of this observation”. See Using F# Option Type in C#

A comment in the answer to the linked question notes:

From section 8.13.1 of the spec: If a property member has both a getter and a setter, and neither is an indexer, then the signatures of both getter and setter must imply the same property type