serialization: add EnumTraits<T> for named enum serialization by kevmeyer · Pull Request #528 · eclipse-score/baselibs

kevmeyer · 2026-08-28T09:11:37Z

Summary

Adds score::common::visitor::EnumTraits<T> — a compile-time trait template that lets an enum type register its enumerator name<->value table — plus a new payload_tags::enum_le tag and enum_serialized_descriptor. visit_as now SFINAE-dispatches to the enum-aware descriptor when EnumTraits<T>::kHasNames is true, and falls back to the previous integer-only memcpy_serialized_descriptor path otherwise (default kHasNames = false, so any enum without a registered trait is unaffected).

The wire format (serialized byte layout) is unchanged — enum_serialized_descriptor uses the same memcpy_serialized<sizeof(T)> payload type as before, only exposing extra compile-time metadata (wire_tag, enum_type) for downstream consumers.

This lets downstream code generators (e.g. FIBEX generation for DLT/CANape tooling) recover proto enumerator names for enum-typed fields, which are otherwise lost once the enum is folded into a plain integral wire type.

Also fixes a latent signed/unsigned dispatch bug introduced by the new enum overloads: std::is_signed<T> is only meaningful for arithmetic types and is always false for enum types (enums are not arithmetic types per the standard), even when the enum's underlying type is signed. Added details::is_signed_enum<T>, which inspects std::underlying_type_t<T> and is only instantiated for enum types, and use it instead of std::is_signed<T> for the enum dispatch — otherwise every enum would incorrectly get wire_tag = unsigned_le.

Testing

Added serializer_visitor.named_enum_wire_format_unchanged unit test in test_serializer_visitor.cpp:

Registers EnumTraits<test::EnumWithNames> and verifies kHasNames defaults to false for enums without a trait (e.g. the pre-existing test::E).
Verifies check_serialized<test::EnumWithNames>() == sizeof(test::EnumWithNames) (wire format unchanged).
Static-asserts the visit_as descriptor for a named enum uses payload_tags::enum_le, exposes the correct enum_type, and dispatches to wire_tag == payload_tags::signed_le for an enum with a signed underlying type (regression check for the signed/unsigned dispatch fix).
Verifies the registered enumerator name<->value table (kEnumerators) is actually retrievable through the trait.

bazel test //score/static_reflection_with_serialization/serialization:unit_tests — all pass (serializer_ut, size_visitor_ut, visitor_type_traits_ut).

github-actions · 2026-08-28T09:55:06Z

Documentation preview for this pull request is available at:
pr-528: https://eclipse-score.github.io/baselibs/pr-528/

Add score::common::visitor::EnumTraits<T> trait template plus a new payload_tags::enum_le tag and enum_serialized_descriptor. visit_as SFINAE-dispatches to the enum-aware descriptor when EnumTraits<T>::kHasNames is true, otherwise falls back to the previous integer-only behavior. The wire format (byte layout) is unchanged. This lets downstream code generators (e.g. FIBEX generation for DLT/ CANape tooling) recover enumerator names for proto-derived enums by specializing EnumTraits<T>, without any change for enums that don't register a trait. Also fixes signed-enum detection in the new visit_as overloads: std::is_signed<T> is only meaningful for arithmetic types and is always false for enum types (enums are not arithmetic types per the standard), even when their underlying type is signed. Added details::is_signed_enum<T>, which inspects std::underlying_type_t<T> and is only instantiated for enum types, and use it in place of std::is_signed<T> for the enum dispatch. Added unit test coverage in test_serializer_visitor.cpp: - Verifies EnumTraits<T>::kHasNames defaults to false for enums without a registered trait, and true once specialized. - Verifies the serialized wire size is unchanged for a named enum. - Verifies visit_as dispatches to the enum_le payload tag and exposes the correct enum_type and wire_tag (signed_le for enums with a signed underlying type). - Verifies the registered enumerator name<->value table (kEnumerators) is actually retrievable through the trait. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

4og · 2026-08-28T13:54:43Z

@rmaddikery, @hoppe-and-dreams, please check this PR

hoppe-and-dreams

Thanks for the contribution. Looks well in general. I added some minor findings to please address.

hoppe-and-dreams · 2026-08-31T06:18:50Z


 }  // namespace payload_tags

+// One value <-> name mapping of an enumerator, as declared in the originating proto `enum`.


This comment is misleading. This library is independent from proto. This might be you use case, but it is generally not correct to use this as the only use case of this library.

hoppe-and-dreams · 2026-08-31T06:21:35Z

+//
+// The primary (unspecialized) template intentionally carries no names: any C++ enum works with
+// the existing signed/unsigned integral serialization path unless a specialization of this trait
+// is provided for it. Code generators are expected to emit a specialization for every proto


Again, the whole proto description is not relevant for this library in S-CORE.

The general comment, about the difference between how "specialized" enums vs regular enums are handle suffices.

hoppe-and-dreams · 2026-08-31T06:24:38Z

+// that downstream consumers (e.g. the FIBEX generator's visitor) can look up EnumTraits<T> and
+// recover the enumerator name<->value table. This does not change the serialized byte layout.
+template <typename Tag, typename T>
+// coverity[autosar_cpp14_a11_0_2_violation]


No coverity in S-CORE repository

hoppe-and-dreams · 2026-08-31T06:26:38Z

+    typename T,
+    std::enable_if_t<(std::is_enum<T>::value) && (details::is_signed_enum<T>::value) && (EnumTraits<T>::kHasNames),
+                     std::int32_t> = 0>
+/// \brief Dispatches a signed enum with a registered EnumTraits<T> specialization to the


Usually the comments go before the template< ... > line

hoppe-and-dreams · 2026-08-31T06:26:46Z

+// coverity[autosar_cpp14_m3_2_3_violation : FALSE]
+// coverity[autosar_cpp14_a2_10_4_violation : FALSE]


Again, coverity

hoppe-and-dreams · 2026-08-31T06:28:05Z

+///         unsigned integer representation as memcpy_serialized_descriptor, but additionally
+///         exposing T so that EnumTraits<T> can be looked up by downstream consumers.
+// This is false positive, Overload signatures are different.
+// coverity[autosar_cpp14_m3_2_3_violation : FALSE]


More coverity here

Its fine for documentation of intent w.r.t to suppression until a new tool is introduced where the suppression could be removed OR carried over.

hoppe-and-dreams · 2026-08-31T06:31:16Z

+    RecordProperty("ParentRequirement", "SCR-1633893");
+    RecordProperty("ASIL", "B");
+    RecordProperty("Description",
+                   "Check that registering EnumTraits<T> for an enum type only adds compile-time "
+                   "enumerator name metadata and does not alter its serialized wire representation.");
+    RecordProperty("TestingTechnique", "Requirements-based test");
+    RecordProperty("DerivationTechnique", "requirements-analysis");  // requirements


None of this makes sense in S-CORE, does it @4og ?

It actually does, but the values are slightly different in S-CORE. The best is to see them in this markdown: https://github.com/eclipse-score/baselibs/blob/main/.agents/skills/requirements-management-skill/test-to-requirement-linking.md

We currently have an epic to clean all of this up across the codebase: #522

For this PR, I would just make the Description, TestingTechnique, DerivationTechnique aligned with the S-CORE guidelines (basically, only the TestingTechnique needs to be turned to ("TestType", "requirements-based"). For the rest I don't care for now.

kevmeyer requested review from 4og, antonkri, arkjedrz, fbaeuerle and pawelrutkaq as code owners August 28, 2026 09:11

github-project-automation Bot added this to BAS - Baselibs FT Aug 28, 2026

github-project-automation Bot moved this to In Progress in BAS - Baselibs FT Aug 28, 2026

kevmeyer deployed to workflow-approval August 28, 2026 09:11 — with GitHub Actions Active

github-actions Bot added comp-static_reflection_with_serialization Related to score/static_reflection_with_serialization c++ C++ code labels Aug 28, 2026

kevmeyer force-pushed the kevinmemeyer/main/enum-traits-support branch from e8515eb to 02a50e1 Compare August 28, 2026 10:12

kevmeyer deployed to workflow-approval August 28, 2026 10:13 — with GitHub Actions Active

hoppe-and-dreams reviewed Aug 31, 2026

View reviewed changes

Navigation Menu

serialization: add EnumTraits<T> for named enum serialization - #528

serialization: add EnumTraits<T> for named enum serialization#528
kevmeyer wants to merge 1 commit into
eclipse-score:mainfrom
kevmeyer:kevinmemeyer/main/enum-traits-support

kevmeyer commented Aug 28, 2026

Uh oh!

github-actions Bot commented Aug 28, 2026

Uh oh!

4og commented Aug 28, 2026

Uh oh!

hoppe-and-dreams left a comment

Uh oh!

hoppe-and-dreams Aug 31, 2026

Uh oh!

hoppe-and-dreams Aug 31, 2026

Uh oh!

hoppe-and-dreams Aug 31, 2026

Uh oh!

hoppe-and-dreams Aug 31, 2026

Uh oh!

hoppe-and-dreams Aug 31, 2026

Uh oh!

hoppe-and-dreams Aug 31, 2026

Uh oh!

rmaddikery Sep 1, 2026

Uh oh!

hoppe-and-dreams Aug 31, 2026

Uh oh!

4og Aug 31, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants


		} // namespace payload_tags

		// One value <-> name mapping of an enumerator, as declared in the originating proto `enum`.

		// coverity[autosar_cpp14_m3_2_3_violation : FALSE]
		// coverity[autosar_cpp14_a2_10_4_violation : FALSE]

Conversation

kevmeyer commented Aug 28, 2026

Summary

Testing

Uh oh!

github-actions Bot commented Aug 28, 2026

Uh oh!

4og commented Aug 28, 2026

Uh oh!

hoppe-and-dreams left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants