Remove [FixedAddressValueType].

[teo-tsirpanis]

It has no effect for this use case and blocks the assembly from being loaded in unloadable AssemblyLoadContexts.

Fun fact: These were the only uses of this attribute on GitHub outside of .NET itself. 😅

[glenn-slayden]

These were the only uses of this attribute on GitHub outside of .NET itself...

Really? I use FixedAddressValueTypeAttribute all the time, typically for read-only reference tables or data that will be referenced for the duration/lifetime of the CLR session.

Obviously, GC and managed memory are the cat's meow, but for some cases, it's nice to have a permanent nint value which lets you immediately indirect into (pointer arithmetic!)--and fetch into--the exact piece data with no muss or fuss. Grab-it-and-go? I like to think, although some purists would probably think of it as more of a smash 'n grab?

[teo-tsirpanis]

read-only reference tables or data that will be referenced for the duration/lifetime of the CLR session

There is a better way to do this. Consider this code:

using System;

public class C
{
    public ReadOnlySpan<byte> Data => new byte[3] { 1, 2, 3 };
}

It might seem that accessing Data will allocate a byte array but it won't. The span will point directly to the data section of the assembly.

This has worked with byte-sized values since quite some time and starting with dotnet/roslyn#61414 it can emit optimized code for arrays of bigger primitives.