Allow using rfl::AddTagsToVariants with rfl::Generic and same-name-different-namespace structs

[jmcken8]

The first one was the main reason for this change; the second one (ex: a::Message and b::Message) was a positive result of applying the change. Added tests for both.

The issue was that both std::vector<Generic> and rfl::Object<Generic> were getting the type name 'Generic', which caused a compiler error due to duplicated field names.

This will slightly increase the size of serialized formats when using the AddTagsToVariants, but it seems like a worthwhile change.

[gemini-code-assist]

Summary of Changes

Hello @jmcken8, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request enhances the rfl::AddTagsToVariants functionality by ensuring that fully qualified type names are used as tags, which resolves previous issues with rfl::Generic types causing name collisions and allows for proper serialization of same-name structs across different namespaces. This change slightly increases the size of serialized formats but significantly improves robustness and flexibility.

Highlights

• Core Logic Change: Modified rfl::parsing::VariantAlternativeWrapper::make_tag to no longer strip namespaces from type names when generating tags for rfl::AddTagsToVariants. This ensures that fully qualified type names are used as tags, resolving issues with name collisions.
• Support for rfl::Generic in Variants: Enabled the correct serialization and deserialization of rfl::Generic types when they are part of a rfl::Variant or std::variant and rfl::AddTagsToVariants is applied. Previously, std::vector<Generic> and rfl::Object<Generic> would both be tagged as 'Generic', leading to compiler errors due to duplicate field names.
• Support for Same-Name-Different-Namespace Structs: Improved compatibility for structs that share the same name but reside in different namespaces (e.g., NamespaceA::Message and NamespaceB::Message). These can now be correctly serialized as variant alternatives using their fully qualified names as tags.
• New Test Cases: Added comprehensive test cases in tests/json/test_add_tag_to_variant.cpp to validate the new behavior for rfl::Generic within variants and for structs with identical names in different namespaces.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point in your pull request via creating an issue comment (i.e. comment on the pull request page) using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in issue comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist is currently in preview and may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments to provide feedback.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This pull request modifies rfl::AddTagsToVariants to include full type names with namespaces when generating tags for variant alternatives, resolving issues with rfl::Generic and enabling the use of structs with identical names from different namespaces. The new tests validate these improvements. I've provided suggestions for the new test file to enhance comment clarity, code formatting, and test suite organization.

[bryceschober]

@jmcken8 IMO, the additional size needed for namespaces, which are often nested, means that this change should be of an opt-in variety, as in rfl::AddNamespacedTagsToVariants. There are serialization formats here that are designed for much more space-constrained contexts than JSON.

[jmcken8]

@bryceschober I'm on the fence - AddTagsToVariants already seems like an edge-case issue where it really should just work no matter what's given to it, especially since there are several available ways to add tags explicitly (rfl::TaggedUnion, rfl::Variant with rfl::Field) if a user is worried about the serialized size.

What would you suggest for the rfl::Generic case, i.e. it contains a std::variant with std::vector<rfl::Generic> -> Generic and rfl::Generic<rfl::Generic> -> Generic? Whether or not namespaces are added, I think a change should be made so AddTagsToVariants works with rfl::Generic.

[bryceschober]

@jmcken8 Well I'm not sure how you would make your solution work either. I'm guessing that you've looked at the test failures reported here by now. Looking at GCC 12's test failure, I see:

[ RUN      ] json.test_add_tag_to_variant_with_generic
/home/runner/work/reflect-cpp/reflect-cpp/tests/json/write_and_read.hpp:35: Failure
Expected equality of these values:
  json_string1
    Which is: "{\"test_add_tag_to_variant::APIResult\":{\"result\":{\"std::__cxx11::basic_string<char>\":\"200\"}}}"
  _expected
    Which is: "{\"test_add_tag_to_variant::APIResult\":{\"result\":{\"std::string\":\"200\"}}}"
Test failed on write. Expected:
{"test_add_tag_to_variant::APIResult":{"result":{"std::string":"200"}}}
Got: 
{"test_add_tag_to_variant::APIResult":{"result":{"std::__cxx11::basic_string<char>":"200"}}}

But those are "internal" namespaces, so they're different on for example LLVM 17:

/home/runner/work/reflect-cpp/reflect-cpp/tests/json/write_and_read.hpp:35: Failure
Expected equality of these values:
  json_string1
    Which is: "{\"test_add_tag_to_variant::APIResult\":{\"result\":{\"std::basic_string<char>\":\"200\"}}}"
  _expected
    Which is: "{\"test_add_tag_to_variant::APIResult\":{\"result\":{\"std::string\":\"200\"}}}"
Test failed on write. Expected:
{"test_add_tag_to_variant::APIResult":{"result":{"std::string":"200"}}}
Got: 
{"test_add_tag_to_variant::APIResult":{"result":{"std::basic_string<char>":"200"}}}

My initial guess is that is the original reason for removing the namespace, and I don't know how you would get to where you want from here.

I don't know your use case's requirements, but you do have the opportunity to differentiate your variant members by using the manual tagging suggested in the docs for variants:

namespace Result {
struct Message {
  using Tag = rfl::Literal<"Result::Message">;
  std::string result;
};
}  // namespace Result

namespace Error {
struct Message {
  using Tag = rfl::Literal<"Error::Message">;
  std::string error;
  int error_id;
};
};  // namespace Error

... or whatever naming method of manual disambiguation you'd like.

I also wonder what would happen if you used more fully qualified aliases for the struct types:

using ResultMessage = Result::Message;
using ErrorMessage = Result::Message;
using Messages = std::variant<ResultMessage, ErrorMessage>;

If that method works, then at least you wouldn't be manually keeping string name around that could get out of sync over time.