LDM Notes for May 28th, 2025

[333fred]

https://github.com/dotnet/csharplang/blob/main/meetings/2025/LDM-2025-05-28.md

Agenda

• Nominal Type Unions

[orthoxerox]

I'm a bit sad to see this happen again, but perhaps it might make sense to push forward with closed enums (#9011) and closed type hierarchies (I can't believe there's no proposal for this one) before tackling ad-hoc unions and optimized struct unions? A big universal proposal that supports every conceivable use case might be simply too complicated to pass through LDM's eye of the needle.

[jnm2]

A big universal proposal that supports every conceivable use case might be simply too complicated to pass through LDM's eye of the needle.

Nailed it.

[MgSam]

Given the issues- I just don't see how this feature can ever land without extensive runtime support. I understand the runtime team has no bandwidth for the foreseeable future but I think this feature necessarily has to be gated on the runtime understanding unions.

[TahirAhmadov]

The question is, can it even be made to work with runtime support?

[sab39]

Wouldn't it be basically the same runtime support as Nullable<T> gets? When one of these special struct types gets boxed, it'd box to the contents of the Value field; when unboxing to one of these struct types, it'd populate the Value field. That's definitely oversimplifying but it doesn't seem like it's radically more difficult.