ReportDiagnosticsIfUnsafeMemberAccess(diagnostics, elementField, syntax);

Do we actually access the field explicitly here? I don't think we do. Therefore, it feels strange that we check "safety" rules for the field. #Closed

I think that motivation for reporting use-site was mostly about reporting bad field types/declarations

I guess, my question is whether it was an explicit design decision to check underlying field for "safety".

I guess, my question is whether it was an explicit design decision to check underlying field for "safety".

The general rule we established, I think, is to check user-declared members that the compiler uses. But we don't need to check well-known members.

For example, it was discussed in last LDM for unsafe https://github.com/dotnet/csharplang/blob/main/meetings/2026/LDM-2026-05-13.md#implicit-calls-and-constructor-edge-cases:

When compiler-generated or implicit execution would call a caller-unsafe member, we will either require an unsafe context or produce an error; we will not silently allow the call.

Do we actually access the field explicitly here? I don't think we do.

That's a good point, I haven't realized that. I will remove this check and we can add an open question for inline array fields.

ReportDiagnosticsIfUnsafeMemberAccess(diagnostics, elementField, node);

Similar question here #Closed

// (7,13): error CS9362: 'Buffer._element0' must be used in an unsafe context because it is marked as 'unsafe' or 'extern'

Does generated code use the field explicitly? #Closed