Dmitry

No, I'm working on an alternative that doesn't simply drop `EQUF` tests, but replaces them with different tests.

Here's my benchmarking based on 4 approaches from PR #606 (each commit is a different approach). https://disk.yandex.ru/i/7tKxDT2YG3o-Yw

@pdimov suggested using a struct like this: ``` struct serialize_options { string_view pinf; string_view ninf; string_view nan; }; ``` to control serializer's behaviour. Usage would be something like ``` os

> I don't get it, why does this need its own streambuf? It doesn't. Vinnie previously directed me at using an output streambuf and I've made one. Then using a...

I've been investigating a different thing and built a 32-bit mingw version. That second problem is indeed related to 32 bit address model.

I've got these numbers: ``` 2147483647 == V 2147483648 == static_cast(V) 2147483647 == static_cast(V) 2147483648 == static_cast(static_cast(V)) 2147483647 == jv.to_number() 2147483647 == static_cast(jv.to_number()) ``` Somehow `to_number` is more precise than...

Is there a way I could test a PR on Xcode 7.3 or 8.3?

I mis-named the issue. The problem isn't just with strings, it's also present with objects and array. There are 2 issues: 1. `operator==` will silently allocate when a value is...

How about this: https://github.com/boostorg/json/pull/651/commits/f62332e46d6b571943009797c537ee1baa8723f2

Is there a way to compare against initializer list without explicitly spelling out `std::initializer_list`?