0

While browsing the implementation of standard library headers in Visual Studio with C++ 20, I came across the type __int64, it looked like a built-in type and I couldn't go to its definition. I googled it and found this article by Microsoft. Apparently the types __int32, __int64 etc are Microsoft-specific built-in types.

I wondered why Microsoft uses these types instead of the non-Microsoft-specific int32_t, int64_t types, shouldn't they achieve the same thing? When I went to the definitions of those, they were just typedefs for types like int and long long, but I assume the implementation still guarantees that types like int32_t have the amount of bits you'd expect. If these types/typedefs are good enough and reliable for the normal user, why does Microsoft use their own ones? If there's an advantage to using these types, why didn't they typedef int32_t etc as __int32 etc? Is there a situation where I might want to use types like __int32 over int32_t?

phuclv
  • 37,963
  • 15
  • 156
  • 475
Tajoshu
  • 47
  • 10
  • 1
    Historic reasons, I’m sure. These standardization of int64_t and int33_t are relatively recent. ( C++11) though I do look forward to seeing if someone has an in-depth explanation. – Taekahn Oct 08 '22 at 00:58
  • Does this answer your question? [Why do so many libraries define their own fixed width integers?](https://stackoverflow.com/questions/70082906/why-do-so-many-libraries-define-their-own-fixed-width-integers) – phuclv Oct 08 '22 at 01:18

1 Answers1

2

Because MSVC existed long before C++11 was introduced. For things that need a fixed size obviously they have to user their own internal type. That's why those have the __ prefix because the standard says that names beginning with double underscores are reserved in the global namespace

That's also the reason why lots of libraries define their own fixed-width types or even C and C++ keywords, for exampe gstreamer's guint32, gint..., OpenCL's cl_int and cl_uint..., Qt's quint32, quint64..., boost's boost::uint32_t, zlib's z_off64_t... because only decades after their advent the standard stdint.h would come into existence

phuclv
  • 37,963
  • 15
  • 156
  • 475
  • Minor point: names that **contain** (not just begin with) two consecutive underscores are reserved. That's a remnant of cfront, which encoded member function names with two underscores between the name of the class and the name of the function. – Pete Becker Oct 08 '22 at 10:23