windows uses long int as if in a 32-bit machine while in 64-bit

Question

I'm using 10 64-bit. when running the following code for example, I get an overflow and a negative number is printed, while on a linux 64-machine I get the actual number printed:

#include <stdio.h>

int main() {
    long int a = 3845354610;
    printf("Hello, World!%ld\n", a);
    scanf("%ld", &a);
    return 0;
}

The output on windows is:

Hello, World!-449612686

and on linux it's the right output.

I tried using the --build-64bit flag in the CMake options in CLion. I'm using mingw-64 (D:\MinGW\mingw-w64\x86_64-8.1.0-posix-seh-rt_v6-rev0\mingw64).

I tried using the cmd to compile:

gcc -Wall -Wextra -Wvla -std=c99 main.c -o ma.exe

but the output was the same. How can this be fixed?

@4386427 although this is a simple example, in places where I need to recieve this number as an input, I won't be able to place that "L" at the end. — Ariel Yael, Aug 12 '21 at 11:59
@4386427 yes, that was the original problem, and it doesn't work — Ariel Yael, Aug 12 '21 at 12:02
@WeatherVane and the question remains: why does a 64bit machine use 32bit long, although the internet says it should be 64 — Ariel Yael, Aug 12 '21 at 12:04
@ArielYael well, that's it then.... 4x8 = 32 bits can't store 3845354610 — Support Ukraine, Aug 12 '21 at 12:05
@ArielYael Where does " the internet says it should be 64" ? — Support Ukraine, Aug 12 '21 at 12:06
@WeatherVane `printf ("%d %d", sizeof(long), sizeof(int));` gives me `4` for both of them — Ariel Yael, Aug 12 '21 at 12:06
MS treats the 64-bits as applying to *pointers*. That's the difference. — Weather Vane, Aug 12 '21 at 12:07
@4386427 about your question please see the comments on eerorika's answer — Ariel Yael, Aug 12 '21 at 12:08
The 32-bit MSVC uses 32-bit pointer, the 64-bit MSVC used 64-bit pointers (same with GCC). But MSVC stays with 32-bit `int ` and `long`. It is allowed to. — Weather Vane, Aug 12 '21 at 12:09
If you are talking about https://en.wikibooks.org/wiki/C_Programming/limits.h then you are not reading the table correct. It says that `long int` must be at least 32 bits. It is **NOT** saying 64 bits. — Support Ukraine, Aug 12 '21 at 12:09
@ArielYael **ANSI standard minimum** is the rigth most column. My guess is that you look at **Typical value** — Support Ukraine, Aug 12 '21 at 12:10
For these reasons I never use `long`. I have never had any need for it. But on a 16-bit embedded system, you can get the 16, 32, and 64-bit integers. In recent times `int32_t` and `int64_t` have been used if you need a definite size. — Weather Vane, Aug 12 '21 at 12:10
Anyway... if the size of an integer is important use `intN_t` — Support Ukraine, Aug 12 '21 at 12:12
@WeatherVane OK. Maybe I didn't understand what you mean by `But on a 16-bit embedded system, you can get the 16, 32, and 64-bit integers.` I thought you were referring to the size of `long`. — eerorika, Aug 12 '21 at 12:16

score 2 · Accepted Answer · answered Aug 12 '21 at 11:57

2

How can this be fixed?

By using long long or std::int64_t. long is required / guaranteed to be at least 32 bits, and that's the size of long on (64 bit) windows.

answered Aug 12 '21 at 11:57

eerorika

232,697
12
197
326

why then [this](https://en.wikibooks.org/wiki/C_Programming/limits.h) site states that on 64-bit the size should be much larger? – Ariel Yael Aug 12 '21 at 12:01
1

@ArielYael It doesn't state such thing. It says: `LONG_MAX - ANSI standard maximum magnitude value - ≥+2,147,483,647` What is unclear about that? – eerorika Aug 12 '21 at 12:04
@ArielYael use the table here and look https://en.cppreference.com/w/cpp/language/types _"Win64 API - LLP64 or 4/4/8 (int and long are 32-bit, pointer is 64-bit)_" – Richard Critten Aug 12 '21 at 12:04
what do they mean by LONG_MAX -> 64 bit compiler ->+9,223,372,036,854,775,807 then? – Ariel Yael Aug 12 '21 at 12:05
@ArielYael Read the title of the column: "**typical** value". "Typical" doesn't mean "always" or "guaranteed" or "required". The author of that page evidently doesn't consider 64 bit Window's implementation typical. – eerorika Aug 12 '21 at 12:06
@eerorika Ok, in that case, I would like to know how to change it to be that – Ariel Yael Aug 12 '21 at 12:07
@ArielYael different platforms have different sizes - MS-Windows need to remain backward compatible with the 32-bit API so kept `long` as 32-bits. – Richard Critten Aug 12 '21 at 12:08
@ArielYael You cannot choose what size `long` is in your language implementation. You could choose to use a language implementation where `long` is 64 bit by not using windows. But it's generally better to not rely on `long` having a particular size allowing your program to be portable. – eerorika Aug 12 '21 at 12:08
@RichardCritten I understand, thank you – Ariel Yael Aug 12 '21 at 12:09
To avoid this type of problem in future, consider adding asserts to your code that confirm your assumptions about the size of your data types. – Tim Randall Aug 12 '21 at 13:52

windows uses long int as if in a 32-bit machine while in 64-bit

1 Answers1