This project is mostly a proof of concept, built on the idea that something we need and want could have been implemented many years ago. If you're curious about the details of how it works, you can find them in this DEV article. Many parts were made just for fun, so they may look interesting or weird. If you want something similar in production, use glaze instead.

It parses C++ source code for special attributes. In the most straightforward situation, you only need to mark an object by [[er::reflect]] attribute. All other work will be done by the code generation tool and reflection library.

The main idea is to use kinda dynamic typing for some type agnostic operations, like copying or getting the name of a type.
It makes it possible to determine the type in runtime and do the right job - print, serialize/deserialize.

• Linux, MacOS and Windows, x86 and ARM support
• translate enums to string and vice versa
• invoke methods
• support of stl containers like std::vector, std::list, std::map, etc.
• smart pointers support
• native serialization directly to an object and without third parties for:
  • JSON
  • YAML 1.2 even with anchors, but keep in mind that variables behind anchors have to be the same type.
  • binary with Variable Length Quantity to reduce number of bytes
• debug printing
• understandable errors

Look at Installation guide and install the solution.

Then define your object and use [[er::reflect]]:

class [[er::reflect]] Object {
 public:
  std::string field_str;
  int field_int;

  std::vector<int> field_vector;
}

And serialize/deserialize it in one shot:

#include "generated/reflection.h"
#include "er/serialization/json.h"

using namespace serialization;

...
auto str = json::to_string<Object>(obj).unwrap();
obj = json::from_string<Object>(str).unwrap();
...

For more details see How To Use.

The repository includes benchmarks folder, feel free to check it on your own hardware.

JSON is faster than nlohmann json. Serialization is the same fast as rapid json, deserialization is a little faster with simdjson parser and more than twice slower without.

YAML is blazingly faster than yaml-cpp if I did the benchmark right.

Note: Other libraries do not always convert string-represented values to int, float, or bool and don't create instances of std::string until you call something like .get<int>(). Easy Reflection, on the other hand, provides ready-made objects with all values within.

The ratio of the content length in bytes