[Serialization] [Feature] Add experimental C++26 reflection support by ujjwal-shekhar · Pull Request #6859 · TheHPXProject/hpx

ujjwal-shekhar · 2026-01-24T14:53:11Z

This PR introduces support for C++26 reflection-based serialization (based on the P2996 proposal). It adds a new reflection-based dispatch mechanism in access.hpp that allows for the automatic serialization of user-defined structs without requiring manual serialize functions. The code generation pertaining to (semi-)automatic polymorphic type registration, serialization lambdas (with and without captures) and compile-time compression of qualified_name_of will be done in the next PR(s).

Proposed Changes

Added HPX_SERIALIZATION_WITH_ALLOW_AUTO_GENERATE to allow users to toggle reflection codegen.
Automatically iterates over bases and data members in refl_serialize_impl.hpp using std::meta.
Utilizes std::meta::access_context::unchecked() and compile-time pointer arithmetic (offset_of) to serialize private members without requiring friend declarations.
Added a qualified_name_of utility to generate compiler-agnostic type identifiers by walking the scope tree. This generates portable and unique names for each type*
Integrated the reflection path into the main serialize dispatch. Manual serialize members and ADL handlers still take precedence, ensuring full backward compatibility.
When reflection-based serialization is enabled, struct_serialization is superseded, as the reflection branch natively handles brace-initializable structs.
Added HPX_POLYMORPHIC_AUTO_REGISTER macro to enable polymorphic type registration.

Any background context you want to provide?

Current HPX serialization requires users to manually list every member in a serialize function, leading to high maintenance overhead and error-prone edits. Encapsulation is also a hurdle, as non-intrusive methods cannot access private data easily.

Goals:

Adding/removing members of a class requires zero changes to serialization code.
Generally, eliminate the need for classes to be "aware" of their serialization logic.
Build on existing HPX/Boost infrastructure without replacing it.

Checklist

Not all points below apply to all pull requests.

I have added a new feature and have added tests to go along with it.
I have fixed a bug and have added a regression test.
I have added a test using random numbers; I have made sure it uses a seed, and that random numbers generated are valid inputs for the tests.

StellarBot · 2026-01-24T14:55:03Z

Can one of the admins verify this patch?

codacy-production · 2026-01-24T15:56:26Z

Coverage summary from Codacy

See diff coverage on Codacy

Coverage variation	Diff coverage
Report missing for `cc3770e`¹	✅ 0.00%

Coverage variation details

	Coverable lines	Covered lines	Coverage
Common ancestor commit (`cc3770e`)	Report Missing	Report Missing	Report Missing
Head commit (`a62e2f1`)	131434	51	0.04%

Coverage variation is the difference between the coverage for the head and common ancestor commits of the pull request branch: <coverage of head commit> - <coverage of common ancestor commit>

Diff coverage details

	Coverable lines	Covered lines	Diff coverage
Pull request (#6859)	132	0	0.00%

Diff coverage is the percentage of lines that are covered by tests out of the coverable lines that the pull request added or modified: <covered lines added or modified>/<coverable lines added or modified> * 100%

See your quality gate settings Change summary preferences

Codacy didn't receive coverage data for the commit, or there was an error processing the received data. Check your integration for errors and validate that your coverage setup is correct. ↩

hkaiser

@ujjwal-shekhar Sorry for spamming with a long list of comments, I went over the PR from a formal standpoint for now. I have not done a deep review of the implementation yet.

Generally, thank you so much for working on this! This is marvelous.

ujjwal-shekhar · 2026-01-24T16:32:20Z

@ujjwal-shekhar Sorry for spamming with a long list of comments, I went over the PR from a formal standpoint for now. I have not done a deep review of the implementation yet.

Generally, thank you so much for working on this! This is marvelous.

Thank you! It is a really interesting feature indeed. No worries about the long list; I'll go through the comments soon and green out the failing checks as well.

hkaiser · 2026-01-26T13:25:07Z

+    template <>
+    struct register_types<>
+    {
+        static void ensure_instantiated() {}


I think these functions can be constexpr and noexcept.

I am considering shifting ensure_instantiated and other improvements to the core auto-serialization facility to a future PR (which will also include a DFS of polymorphic types for semi-automation of polymorphic types, compile-time checks to see if the user missed a register; etc).

hkaiser

This starts looking really nice! Just some mostly minor comments below...

hkaiser · 2026-02-15T15:59:54Z

inspect and the cmake file checker are complaining about problems, please have a look. The other CircleCI problem are known and unrelated.

ujjwal-shekhar · 2026-02-16T17:33:56Z

Thank you for the comments @hkaiser !

I have made some changes addressing most of the comments. Before wrapping up the rest, I will focus on fixing the new CI script with cpp26 reflection (as I expect there to be some bugs when testing the full suite under enabled reflection changes :p)

hkaiser

Just some food for thought: what can be made consteval instead of being constexpr?

ujjwal-shekhar · 2026-03-03T06:42:33Z

Just some food for thought: what can be made consteval instead of being constexpr?

At a glance, I am inclined to believe that anything that must be evaluated within the std::meta context should be constevalable. By extension, it feels like a lot of this could be consteval. I will experiment with this:)

ujjwal-shekhar · 2026-03-08T07:11:47Z

@hkaiser, In the CI job with reflection enabled, would you recommend we run the full suite or just the serialization tests? The full suite would turn it into a long job so I have limited it to just the serialization tests for now.

codacy-production · 2026-03-08T12:19:57Z

Coverage summary from Codacy

See diff coverage on Codacy

Coverage variation	Diff coverage
Report missing for `358fef1`¹	✅ 0.00%

Coverage variation details

	Coverable lines	Covered lines	Coverage
Common ancestor commit (`358fef1`)	Report Missing	Report Missing	Report Missing
Head commit (`ce9670f`)	131434	51	0.04%

Coverage variation is the difference between the coverage for the head and common ancestor commits of the pull request branch: <coverage of head commit> - <coverage of common ancestor commit>

Diff coverage details

	Coverable lines	Covered lines	Diff coverage
Pull request (#6859)	131	0	0.00%

Diff coverage is the percentage of lines that are covered by tests out of the coverable lines that the pull request added or modified: <covered lines added or modified>/<coverable lines added or modified> * 100%

See your quality gate settings Change summary preferences

Codacy didn't receive coverage data for the commit, or there was an error processing the received data. Check your integration for errors and validate that your coverage setup is correct. ↩

hkaiser · 2026-03-08T14:49:14Z

@hkaiser, In the CI job with reflection enabled, would you recommend we run the full suite or just the serialization tests? The full suite would turn it into a long job so I have limited it to just the serialization tests for now.

Just the serialization tests for now, I guess. We will soon move this repository to a new github organization where we will have access to an enterprise license. We can re-evaluate the CI situation then.

hkaiser · 2026-03-08T14:55:00Z

@ujjwal-shekhar Overall, this PR seems to be almost complete - excellent work!

ujjwal-shekhar · 2026-03-08T18:25:32Z

@ujjwal-shekhar Overall, this PR seems to be almost complete - excellent work!

@hkaiser Thanks! I'm really happy with how the reflection logic is holding up. Once we've addressed the remaining comments, I'll move this from a draft to an open PR. Super excited to get this merged!

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

… code, macro can be improved, test CRTP serialization Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

…tcast instead Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

…so far) Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

ujjwal-shekhar · 2026-03-12T16:44:44Z

The CMake build system checker reports:
Missing hpx/serialization/traits/is_serializable.hpp in libs/core/serialization/CMakeLists.txt
Could you please add that file?

Done @hkaiser ! Apologies for having missed this, I had accidentally removed this line while adding the new trait file to the CMakeList.

codacy-production · 2026-03-13T01:25:24Z

Coverage summary from Codacy

See diff coverage on Codacy

Coverage variation	Diff coverage
Report missing for `89914d3`¹	✅ 0.00%

Coverage variation details

	Coverable lines	Covered lines	Coverage
Common ancestor commit (`89914d3`)	Report Missing	Report Missing	Report Missing
Head commit (`b095cd4`)	130899	51	0.04%

Coverage variation is the difference between the coverage for the head and common ancestor commits of the pull request branch: <coverage of head commit> - <coverage of common ancestor commit>

Diff coverage details

	Coverable lines	Covered lines	Diff coverage
Pull request (#6859)	149	0	0.00%

Diff coverage is the percentage of lines that are covered by tests out of the coverable lines that the pull request added or modified: <covered lines added or modified>/<coverable lines added or modified> * 100%

See your quality gate settings Change summary preferences

Codacy didn't receive coverage data for the commit, or there was an error processing the received data. Check your integration for errors and validate that your coverage setup is correct. ↩

hkaiser · 2026-03-13T13:15:38Z

@ujjwal-shekhar thank you again for your persistence and effort! This is a major milestone for HPX!

…ialize [Serialization] [Feature] Add experimental C++26 reflection support

ujjwal-shekhar force-pushed the reflect-serialize branch from b1e3990 to 4c26be6 Compare January 24, 2026 15:31

hkaiser added type: enhancement type: compatibility issue category: serialization labels Jan 24, 2026

hkaiser added this to the 2.0.0 milestone Jan 24, 2026

hkaiser reviewed Jan 24, 2026

View reviewed changes

github-advanced-security AI found potential problems Jan 24, 2026

View reviewed changes

Comment thread libs/core/serialization/include/hpx/serialization/input_archive.hpp Fixed

Comment thread libs/core/serialization/include/hpx/serialization/output_archive.hpp Fixed

ujjwal-shekhar force-pushed the reflect-serialize branch from 23c54d8 to 9e8686b Compare January 25, 2026 11:06

hkaiser reviewed Jan 25, 2026

View reviewed changes

hkaiser reviewed Jan 26, 2026

View reviewed changes

ujjwal-shekhar force-pushed the reflect-serialize branch from 6c9ea93 to 5b78342 Compare February 7, 2026 05:43

hkaiser reviewed Feb 15, 2026

View reviewed changes

ujjwal-shekhar force-pushed the reflect-serialize branch from edeb250 to c877a23 Compare February 16, 2026 17:29

hkaiser reviewed Feb 16, 2026

View reviewed changes

Comment thread libs/core/serialization/include/hpx/serialization/detail/refl_serialize_impl.hpp Outdated

hkaiser reviewed Feb 17, 2026

View reviewed changes

Comment thread libs/core/serialization/include/hpx/serialization/detail/refl_serialize_impl.hpp Outdated

Comment thread libs/core/serialization/include/hpx/serialization/input_archive.hpp Fixed

ujjwal-shekhar force-pushed the reflect-serialize branch from 67659aa to 3a172bb Compare March 2, 2026 12:14

hkaiser reviewed Mar 2, 2026

View reviewed changes

ujjwal-shekhar force-pushed the reflect-serialize branch 2 times, most recently from d4d906d to ce9670f Compare March 8, 2026 07:03

ujjwal-shekhar added 24 commits March 12, 2026 16:42

add new test file

60c56b2

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

fmt

7735321

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

polymorphism has started working! some improvements to be done to the…

def35fd

… code, macro can be improved, test CRTP serialization Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

clang fmt

b97ae4f

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

fixing build issues

0daaacf

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

Addressing comments, improving noexceptness of refl_qualified_name_of

019b319

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

Fixing merge conflicts, HPX_CXX_EXPORT conflicts only

bc442f2

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

Addressing comments, removing polymorphic_type_registrar and using bi…

f23315a

…tcast instead Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

Changing to byte*, addressing comments

ec57d31

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

fixing ci workflow script

1e6d905

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

fixing ci workflow script build issue, figuring out issues

8245a37

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

Addressing comments, trying to reduce build time for the CI script

aa38780

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

Making some functions consteval in refl_qualified_name_of

3f798a1

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

Fixing build with brace_initializable removed

1e8b8be

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

inspect issue

26d116c

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

cleanup make_args_list

8d7ab4c

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

Fixing serialization of protected and private bases (maybe WIP, lgtm …

0b8dc96

…so far) Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

Improving traits and ifdef clutter in input/outputarchive.hpp

034dd27

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

adding file

4c7202d

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

fixing inspect, fmt etc

dca49df

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

fixing inspect issues

30d3e9e

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

rename member_from_offset, and relocate. Address comments

e55ed0a

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

addressing comments

967d7f7

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

fix cmakelists issue

b095cd4

Signed-off-by: Ujjwal Shekhar <ujjwal.shekhar@research.iiit.ac.in>

ujjwal-shekhar force-pushed the reflect-serialize branch from f42e937 to b095cd4 Compare March 12, 2026 16:42

hkaiser merged commit c982aff into TheHPXProject:master Mar 13, 2026
111 of 114 checks passed

ArivoliR pushed a commit to ArivoliR/hpx that referenced this pull request Mar 29, 2026

Merge pull request TheHPXProject#6859 from ujjwal-shekhar/reflect-ser…

e7e1fe0

…ialize [Serialization] [Feature] Add experimental C++26 reflection support

Uh oh!

Conversation

ujjwal-shekhar commented Jan 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Proposed Changes

Any background context you want to provide?

Checklist

Uh oh!

StellarBot commented Jan 24, 2026

Uh oh!

codacy-production Bot commented Jan 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Coverage summary from Codacy

Footnotes

Uh oh!

hkaiser left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ujjwal-shekhar commented Jan 24, 2026

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

hkaiser left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

hkaiser commented Feb 15, 2026

Uh oh!

ujjwal-shekhar commented Feb 16, 2026

Uh oh!

Uh oh!

Uh oh!

Uh oh!

hkaiser left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ujjwal-shekhar commented Mar 3, 2026

ujjwal-shekhar commented Jan 24, 2026 •

edited

Loading

codacy-production Bot commented Jan 24, 2026 •

edited

Loading

hkaiser commented Mar 8, 2026 •

edited

Loading