Nullability analysis doesn't quite work for user-defined conversions involved into tuple value conversions

Description

        [Fact]
        public void Test()
        {
            var src = @"
#nullable enable

struct S1
{
    public static implicit operator S1([System.Diagnostics.CodeAnalysis.NotNull] string? x) => throw null!;
}

class Program
{
    static void Test3((string?, int) x)
    {
        (S1, int) s = x;
        x.Item1.ToString();
    } 

    static void Test4(string? x)
    {
        S1 s = x;
        x.ToString();
    } 
}
";
            var comp = CreateCompilation([src, NotNullAttributeDefinition]);
            comp.VerifyDiagnostics(
                // (13,23): warning CS8619: Nullability of reference types in value of type '(string?, int)' doesn't match target type '(S1, int)'.
                //         (S1, int) s = x;
                Diagnostic(ErrorCode.WRN_NullabilityMismatchInAssignment, "x").WithArguments("(string?, int)", "(S1, int)").WithLocation(13, 23),
                // (14,9): warning CS8602: Dereference of a possibly null reference.
                //         x.Item1.ToString();
                Diagnostic(ErrorCode.WRN_NullReferenceReceiver, "x.Item1").WithLocation(14, 9)
                );
        }

Contributor guide

Tech stack

csharp

Domain

backend

Issue type

bug

DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.

4

Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.

over 1 week

Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.

blocked

ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.

clear

Prerequisites

Familiarity with C# nullabilityBasic understanding of Roslyn compiler internalsAbility to build and run tests

Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.

20

Research direction

The issue provides a minimal reproduction test case in the issue body. The fix likely requires modifying the nullable analysis logic for tuple conversions when user defined implicit conversions are involved. Look at the Roslyn source code for nullable diagnostics, particularly the handling of tuple conversions in the nullable walker. The expected behavior is that no warning should be emitted for Test3, as the string? is converted to S1 via implicit operator that has [NotNull] attribute. Test4 already works correctly. Investigate the difference between tuple and non tuple assignments to find where the nullability mismatch check is overly strict for tuples.