dump(): incorrect precision for float array

[yhmo]

What is the issue you have?

I am using nlohmann::json version 0.3.7 in my project, it is a great tool.

But I encounter a small problem: dump a json object which contains float array, seems the precision is incorrect.

Can you provide a small but working code example?

#include <vector>
#include <iostream>

#include "json.hpp"

int main() {
    std::vector<float> float_data = {0.1234567, 4.123456};
    nlohmann::json json;
    json["AAA"] = float_data;
    std::string ss = json.dump();
    std::cout << ss << std::endl;
}

But the output string is:
{"AAA":[0.12345670163631439,4.123456001281738]}

What is the expected behavior?

I suppose the output string should be:
{"AAA":[0.1234567,4.123456]}

Which version of the library did you use?

• latest release version 3.7.0

[nlohmann]

The library uses double as default type for floating-point numbers. Therefore, your numbers are converted to double first and then serialized using as many digits as necessary to roundtrip doubles.

If you want to use float instead, you can override the default type templates this:

#include <iostream>
#include "json.hpp"

int main()
{
    std::vector<float> float_data = {0.1234567, 4.123456};
    nlohmann::basic_json<std::map, std::vector, std::string, bool, std::int64_t, std::uint64_t, float> json;
    json["AAA"] = float_data;
    std::string ss = json.dump();
    std::cout << ss << std::endl;
}

Output:

{"AAA":[0.1234567,4.123456]}

[yhmo]

It works. Thank you!