When using MSVC to compile v12.0.0, many compilation warnings are generated

[esoft-com]

size_t and unsigned __int64 do not match when compiling with x86, resulting in the compiler generating a lot of compilation information

fmt\format.h(2509,32): warning C4244: 'argument': conversion from 'unsigned __int64' to 'size_t', possible loss of data [app.vcxproj]
fmt\format.h(3505): message : see reference to function template instantiation'OutputIt fmt::v12::detail::write_fixed<Char,fmt::v12::detail::fallback_digit_grouping<Char>,OutputIt,fmt::v12::detail::dragonbox::decimal_fp<float>>(OutputIt,const DecimalFP &,int,Char,const fmt::v12::format_specs &,fmt::v12::sign,fmt::v12::locale_ref)' being compiled [app.vcxproj]
         with
         [
             OutputIt=fmt::v12::basic_appender<char>,
             Char=char,
             DecimalFP=fmt::v12::detail::dragonbox::decimal_fp<float>
         ]
fmt\format.h(3649): message : see reference to function template instantiation 'OutputIt fmt::v12::detail::write<Char,fmt::v12::basic_appender<char>,T,0>(OutputIt,T)' being compiled [app.vcxproj]
         with
         [
             OutputIt=fmt::v12::basic_appender<char>,
             Char=char,
             T=float
         ]
fmt\base.h(2541): message : see reference to function template instantiation 'void fmt::v12::detail::default_arg_formatter<char>::operator ()<float,0>(T)' being compiled [app.vcxproj]
         with
         [
             T=float
         ]
fmt\base.h(2541): message : see reference to function template instantiation 'void fmt::v12::detail::default_arg_formatter<char>::operator ()<float,0>(T)' being compiled [app.vcxproj]
         with
         [
             T=float
         ]
fmt\format-inl.h(1460): message : see reference to function template instantiation 'void fmt::v12::basic_format_arg<Context>::visit<fmt::v12::detail::default_arg_formatter<char>>(Visitor &&)const' being compiled [app.vcxproj]
         with
         [
             Context=fmt::v12::context,
             Visitor=fmt::v12::detail::default_arg_formatter<char>
         ]
fmt\format-inl.h(1460): message : see reference to function template instantiation 'void fmt::v12::basic_format_arg<Context>::visit<fmt::v12::detail::default_arg_formatter<char>>(Visitor &&)const' being compiled [app.vcxproj]
         with
         [
             Context=fmt::v12::context,
             Visitor=fmt::v12::detail::default_arg_formatter<char>
         ]

[LiangHuDream]

@esoft-com This warning occurs because on x86 Windows, size_t is a 32-bit unsigned integer, while unsigned __int64 is a 64-bit unsigned integer. Implicitly converting from unsigned __int64 to size_t can cause data loss, so MSVC emits warning C4244.
To resolve this, we should explicitly cast the value to size_t wherever the conversion happens, ideally at the point where the warning is triggered. thank you for checking the code

[esoft-com]

@esoft-com This warning occurs because on x86 Windows, size_t is a 32-bit unsigned integer, while unsigned __int64 is a 64-bit unsigned integer. Implicitly converting from unsigned __int64 to size_t can cause data loss, so MSVC emits warning C4244.出现此警告的原因是，在 x86 Windows 上，size_t 是 32 位无符号整数，而 unsigned __int64 是 64 位无符号整数。从 unsigned __int64 隐式转换为 size_t 可能会导致数据丢失，因此 MSVC 发出警告 C4244。 To resolve this, we should explicitly cast the value to size_t wherever the conversion happens, ideally at the point where the warning is triggered. thank you for checking the code为了解决这个问题，我们应该在转换发生的任何地方明确地将值转换为 size_t，最好是在触发警告的地方。感谢您检查代码

The problematic code is from fmtlib, not from my program

#include <iostream>

#define FMT_HEADER_ONLY
#include "fmt/core.h"
#include "fmt/xchar.h"

int main(int argc, char **argv)
{
    return 0;
}

As long as the fmtlib library is used, there will be a warning

VS2022
ISO C++14
x86

[vitaut]

Fixed in #4572.

[Gfyss]

For anyone else who has encountered this issue with MSVC and ended up here from a google search, you have the following options:

1 - Reduce the logging level for the entire project to "/W2" under Project->Properties->C/C++ General->Warning Level
- Quick and easy, but may mask other, more useful level 3 warnings since it impacts the entire project.

2 - Edit the format.h file and bracket the "offending code" with the following:

#ifdef _M_IX86
#pragma warning(push)
#pragma warning(4:4244)
#endif
<offending code>
#ifdef _M_IX86
#pragma warning(pop)
#endif

Option 2 preserves the int64 warnings if you set the log level to "/W4", but silences them at "/W3" which is the default level for MSVC. It only changes the warning level if the target is x86, and doesn't mask any other level 3 warnings in your project. The down side is that the change will disappear if/when you update your fmt library.

Here's my full work-around starting around line 2536 in the 12.1.0 format.h file:

#ifdef _M_IX86
#pragma warning(push)
#pragma warning(4:4244)
#endif
    return write_padded<Char, align::right>(
        out, specs, to_unsigned(size), [&](iterator it) {
          if (s != sign::none) *it++ = detail::getsign<Char>(s);
          it = write_significand(it, f.significand, significand_size, exp,
                                 decimal_point, grouping);
          return num_zeros > 0 ? detail::fill_n(it, num_zeros, Char('0')) : it;
        });
  }
  // 1234e-6 -> 0.001234
  int num_zeros = -exp;
  if (significand_size == 0 && specs.precision >= 0 &&
      specs.precision < num_zeros) {
    num_zeros = specs.precision;
  }
  bool pointy = num_zeros != 0 || significand_size != 0 || specs.alt();
  size += 1 + (pointy ? 1 : 0) + num_zeros;
  return write_padded<Char, align::right>(
      out, specs, to_unsigned(size), [&](iterator it) {
        if (s != sign::none) *it++ = detail::getsign<Char>(s);
        *it++ = Char('0');
        if (!pointy) return it;
        *it++ = decimal_point;
        it = detail::fill_n(it, num_zeros, Char('0'));
        return write_significand<Char>(it, f.significand, significand_size);
      });
#ifdef _M_IX86
#pragma warning(pop)
#endif

Good luck!