No nullable warning emmited for non-nullable properties on structs.

[KimBrusevold]

Version Used:
Using SDK .Net 8.0.107, x64

Steps to Reproduce:
https://sharplab.io/#v2:CYLg1APgAgTAjAWAFBQMwAJboMLoN7LpGYZQAs6AsgBQCU+hxTAbgIYBO6ALgKYDOXdAF50AOx4B3dAGUu7AK4BjLgBV+XOgG5GTIlDgBOarwEA6NQA8u5gPYBVAA4Oe7OrW1Idutp15c+AELCYpI4ADasfHxqAlrIXkz6Rn6B5jxWto7OrrTuXgC+yIWeJfEoGAIKyjJySqrqyARIiaRwAAzoloJ46ADmPFya6HwDQ8XFyGiYMOGR0Q1ITS2Y7Z3p3X2jw1vjyEA===

Enable nullable in project file
<Nullable>enable</Nullable>

Create one struct and one class:

struct StructTest
{
    public string Text { get; set; }
}

class ClassTest
{
    public string Text { get; set; }
}

Expected Behavior:

A nullable warning that the non-nullable property 'Text' cannot be initialized to null on both the struct and class.

Actual Behavior:

Only the class property emits the error:
"Non-nullable property 'Text' must contain a non-null value when exiting constructor. Consider adding the 'required' modifier or declaring the property as nullable.CS8618"

If we treat this warning as an error, the compiler generates invalid code. By obtaining a reference to the struct property Text, we may get an null reference error, even though we declared the property as non-nullable. The empty initializer for the struct (new StructTest()) which calls default(StructTest) sets null in the non-nullable fields

[RikkiGibson]

Unfortunately, structs have known/expected gaps with null safety. There have been various proposals to address this over the years such as dotnet/csharplang#146.

But, the struct scenario you have shown is not expected to issue any warnings under the current design. One way to work around this and get a certain (limited) degree of nullability checking is to explicitly declare a constructor in struct StructTest. There will still be null safety gaps when the default value is used, however.

[KimBrusevold]

Thanks. I'm not at liberty to say anything about what is expected and not.

I got around this by creating properties and marking them as required. The compiler recognized that de default "empty" constructor is invalid (which is a good thing).
The nullable feature is such a good thing, and seems so close to being good enough.

I really want to be confident in the compiler checking the nullable stuff for me.