Greg Marr

@nlohmann Does this still need work?

Rather than rewrite all of these algorithms which have been in production for many years, and potentially break the use on a user-defined type that specializes `std::front_insert_iterator` or `std::insert_iterator`, I...

I'm not sure how this function can work. We don't know what type `number_float_t` is so we can't create a proper format string for `snprintf` without casting it to something...

> Yes, I think casting to `double` is the best thing we can do here. > > But any idea on the different behavior between the old approach (snprintf) and...

@nlohmann I think you mean #4758

Okay, so that's the "cast to double" solution rather than the "use iostreams" solution that @t-b suggested. I think that since we don't know what type `x` is, and whether...

From the README: If you are using [NuGet](https://www.nuget.org/), you can use the package [nlohmann.json](https://www.nuget.org/packages/nlohmann.json/). Please check https://github.com/nlohmann/json/issues/1132#issuecomment-452250255 on how to use the package. Please file issues [here](https://github.com/hnkb/nlohmann-json-nuget/issues).

@pavelschon This comment wasn't about how to use nuget, but about who maintains the nuget package. The link for filing issues is in the repo where that package is maintained.

I think maybe you should add `inline` before the definition. I've never tried that with a out of line defaulted function before.

Oh, yeah, if there wasn't a destructor there before and you add one, it suppresses the other of the special 5.