Added: LDM Notes for July 16, 2018

[agocke]

https://github.com/dotnet/csharplang/blob/master/meetings/2018/LDM-2018-07-16.md

Agenda:

1. Null-coalescing assignment
   1. User-defined operators
   2. Unconstrained type parameters
   3. Throw expression the right-hand side
2. Nullable await
3. Nullable pointer access
4. Non-nullable reference types feature flag follow-up

[Logerfo]

This is a hole in null-coalescing as well. Both features should be allowed on unconstrained type parameters.

I don't get it. Is this a typo, missing the negation (both features should NOT be allowed), or the null coalescing behavior will change in order for both features be allowed on unconstrained type parameters?

[gafter]

This is not a typo. Both features should work for unconstrained type parameters.

[Joe4evr]

or the null coalescing behavior will change in order for both features be allowed on unconstrained type parameters?

Correct, see #1720.

[bbarry]

Let's favor the semantics of assignment. The return type should be the type of x.

That seems backwards given that the previous conclusion suggested the semantics should be the same as (x ?? (x = y)). Am I reading something wrong?

[333fred]

@bbarry, the type of (x ?? (x = y)) is also the type of x. The type of the left is just X, and the type of the right is the type of the x = y assignment, which is also just X.

[333fred]

More precisely, given a ?? (a = a0), where A is the type of a and is a nullable type, and A0 is the underlying nullable type, and that the type of a0 is A0, we have:
a ?? (a = a0) (substitute types) -> A ?? (A = A0) (simplify assignment) -> A ?? A (simplify coalesce) -> A
While we accept that there could be a few use cases for the assignment to yield A0, we feel that this would break the substitution property of assignment, where you can always substitute the left hand side of an assignment for the entire assignment without changing the containing expression (ie, the same overload would be chosen).

[paulomorgado]

The spec also states that the semantics are equivalent to x = x ?? y, but that should instead be (x ?? (x = y)). If x is null, the assignment is never performed.

Shouldn't it be If x is not null, the assignment is never performed. instead?

[svick]

@paulomorgado You're not the first person to notice that, see #1856.