Added: LDM Notes for 07/01/2020

[333fred]

https://github.com/dotnet/csharplang/blob/master/meetings/2020/LDM-2020-07-01.md

Agenda:

1. Non-defaultable struct types and parameterless struct constructors
2. Confirming unspeakable Clone method implications

Note: This will be the last Wednesday LDM for a while. Despite coronavirus, we're going to proceed with our normal summer slowdown, and hold meetings only on Mondays for the next couple of months to encourage people to take some time off. So the notes aren't missing, they just don't exist :).

[Joe4evr]

One thing that is clear from discussion is that non-defaultable struct types will need to have some standardized form of checking whether they are the default instance. ImmutableArray<T>, for example, has an IsDefault property, as do most of the existing struct types in Roslyn that cannot be used when default.

I would like to throw in my suggestion of one or two attributes that would inform the flow-analysis. Even if a more standard form of default checking gets implemented, having these to assist for APIs that already exist and are used as such today will help a lot.

[ufcpp]

Can't structs participate in with expressions in C# 9.0 milestone? struct's memberwise copy doesn't need a Clone method.

[333fred]

Can't structs participate in with expressions in C# 9.0 milestone? struct's memberwise copy doesn't need a Clone method.

The Clone things we talked about were specific to record types only. C# 9 will not have struct records.

[ufcpp]

struct can have init-only properties in C# 9.0. with needs only memberwise copy and init-only properties. Nevertheless, won't with expressions for value types be implemented until struct records?

[333fred]

struct can have init-only properties in C# 9.0. with needs only memberwise copy and init-only properties. Nevertheless, won't with expressions for value types be implemented until struct records?

I'm unsure what you're asking for here. We didn't discuss anything about struct records or structs in with expressions in this meeting.

[ufcpp]

Are with expressions completely part of records and not considered an orthogonal feature like init-only properties? There is a demand for non-destructive mutation for immutable value types but it's not discussed before struct records?

[ufcpp]

record is something like a reference type that have struct semantics. The unspeakable Clone methods discussed in this LDM are very similar to struct's implicit copy.

[ufcpp]

Therefore whatever future method we add that will allow a regular class type to participate in a with expression will likely have to emit this method as well.

Does struct also need a user-overridable Clone method in the future?

[CyrusNajmabadi]

Does struct also need a user-overridable Clone method in the future?

It has not been designed yet.

[ufcpp]

OK. I understand this is out of scope for this LDM but my concern is that regular structs will be deprecated when record structs are added, like anonymous methods after lambda expressions.

[orthoxerox]

@333fred Could you please spell out the name of the month in the LDM notes issue names? It will save everyone from DD/MM countries at least 15 seconds of confusion per LDM.

[CyrusNajmabadi]

OK. I understand this is out of scope for this LDM but my concern is that regular structs will be deprecated when record structs

There is no way that regular structs would be deprecated.

[orthoxerox]

Some older community discussions for structs with custom invalid values:

• [Proposal] Make ?? pattern-based for struct left-hand-side roslyn#4347
• Proposal: user-defined null/default check (non-defaultable value types / nullable-like types) roslyn#15108
• Proposal: nullable-like types #147
• Proposal: Overridable `default` operator #167
• Nullable pattern for custom type #494

[333fred]

@333fred Could you please spell out the name of the month in the LDM notes issue names? It will save everyone from DD/MM countries at least 15 seconds of confusion per LDM.

I'll keep that in mind. It is fully spelled out in the notes themselves.