[jogibear9988]

A TypeDelegator or unfinished TypeBuilder constant is treated as an ldtoken-compatible runtime Type, so strict compilation throws ArgumentException instead of returning the original object. Keep non-runtime Type instances in FEC's existing constants closure. Actual runtime Type values retain token emission and their identity.

One constant-classification condition and a cached concrete runtime Type. The comparison now uses a named RuntimeType field, initialized once, with a source comment explaining why TypeDelegator/TypeBuilder must stay in the closure to preserve identity (review feedback). TypeCode.Object does not distinguish these implementations. Keep the ordinary constant short-circuit order. Shared Standard/Light tests cover identity, repeated constants, null, runtime/generic parameter, byref and array Type values.

This is an independent extraction from #558, based directly on master 750e1e8ea602a7e72a90072ee80b0c1376f5b72d. It has the original fix plus a separate review-clarification commit and does not depend on either of the other standalone fixes. #558 and #559 are unchanged.

Validation

• The new regressions fail against unmodified master and pass with this fix on .NET 8, .NET 10 and .NET Framework 4.7.2 (4 registered regression methods across the applicable input APIs).
• Full Release solution build passes with zero warnings/errors.
• Legacy native suites pass: 1,756 cases on .NET 10, 1,755 on .NET 8, and 1,639 on Framework.
• The complete Framework runner passes 194 TestX + 1,639 legacy cases. en-US culture is used for existing English exception-message assertions.
• On .NET 8/10, the full runner still fails ParityTests.Binary_operators_with_methods_and_nullable_operands_are_the_same_as_system for nullable MultiplyChecked in both input variants. The identical failures reproduce on unmodified master. No other fixture failures were observed; the legacy suites were invoked separately because this existing TestX failure gates the normal runner.

Rebase

Rebased onto freshly fetched upstream master 750e1e8e on November 3, 2026. Both test-runner registration conflicts were resolved by retaining master's reduction regressions alongside this PR's identity regressions. The PR remains two focused commits directly above master. All validation results above were rerun after the rebase.

The earlier performance table used the previous master base and has been removed. Performance measurements have not been rerun against this new base; no timing or allocation claims are made for the rebased branch.

[dadhi]

The type != typeof(object).GetType() check is correct, but it’s not obvious why we compare against that type. Could we make the intent explicit and evaluate it once?

private static readonly Type RuntimeType = typeof(object).GetType();

// Only the real RuntimeType can be emitted via ldtoken; TypeDelegator/TypeBuilder must stay in the closure to preserve identity.

Then use type != RuntimeType in IsClosureBoundConstant. TypeCode won’t distinguish these types: GetTypeCode returns Object for RuntimeType, TypeDelegator, and TypeBuilder.

[jogibear9988]

Implemented in 55dca61. IsClosureBoundConstant now compares against private static readonly Type RuntimeType = typeof(object).GetType(), so the implementation type is evaluated once. The source comment explains why only real runtime Type objects use token emission, while TypeDelegator/TypeBuilder stay in the closure to preserve their identity. Agreed that TypeCode.Object cannot distinguish these implementations.

The existing identity regressions cover TypeDelegator, unfinished TypeBuilder, repeated references, and runtime Type values for both Standard and LightExpression. They pass on .NET 8, .NET 10, and .NET Framework 4.7.2.

Also rebased onto freshly fetched master (750e1e8), preserving both sets of test registrations. Release solution build passes without warnings/errors; legacy suites and Framework runner pass. The previously observed nullable MultiplyChecked parity failures remain in the .NET 8/10 full runners.