'long long' not standard

Hi,

I was doing a bit of reading into the C++03 standard and I noticed that 'long long' is not standard. If it is not standard, why have so many compilers (Especially MSVC) made up a non-standard integral type?

On 32-bit windows, 'unsigned int' happens to be 32-bits. Also, 'unsigned long' happens to be 32-bits. They could have completely eliminated the need for 'long long' if they had made 'long' 64-bits on 32-bit architectures.

Why didn't things turn out this way? I want to be as standards conforming as possible in my code, which means I can't use 'long long'. However, I need access to 64-bit integrals on 32-bit architectures (Windows XP, in this example). Is there anything I can do?

Thanks.

Just use long long. C++0x will standardize it anyway, and you don't have any better options.

By the way, long is a 32-bit type on 64-bit Windows, too. That's the part that's *really* strange. (It's 64 bits on 64-bit Linux.)

Lang Lang is a fabulous pianist. I heard him play the first Tchaikovsky concerto with the Cleveland Orchestra a few seasons ago.

Quote:

---

Originally Posted by Lindley

Just use long long. C++0x will standardize it anyway, and you don't have any better options.

By the way, long is a 32-bit type on 64-bit Windows, too. That's the part that's *really* strange. (It's 64 bits on 64-bit Linux.)

---

On 64-bit windows, long is only 32 bits? What the? That means 'int' must be less than or equal to 32 bits!!

Technically it's a property of the compiler rather than OS:
http://en.wikipedia.org/wiki/64-bit#...ic_data_models

Quote:

---

Originally Posted by 0xC0000005

Lang Lang is a fabulous pianist. I heard him play the first Tchaikovsky concerto with the Cleveland Orchestra a few seasons ago.

---

I think, he's a bit overrated. I also heard him a couple years back and thought he was good but not fabulous. Or maybe it's just that here in Germany there is too much fuzz being made about him...

Quote:

---

Originally Posted by MrDoomMaster

On 64-bit windows, long is only 32 bits? What the?

---

The Old New Thing: Why did the Win64 team choose the LLP64 model?


The forgotten problems of 64-bit programs development

Quote:

---

Originally Posted by 0xC0000005

Lang Lang is a fabulous pianist. I heard him play the first Tchaikovsky concerto with the Cleveland Orchestra a few seasons ago.

---

And Ling Ling was a famous zoo animal.

Anyway, the "bitness" used by a compiler is usually tied to the CPU architecture. However nothing stops a compiler to have integers of any bits and on any platform. In these cases, the compiler would emulate an "n-bit" integer via software, not hardware,

This was the case with old DOS compilers that had 32-bit longs. Even though the Intel architecture at that time was 16-bit, 32-bit longs were implemented by using emulation (if you debugged into the calls that generated the 32-bit longs, you actually could see a 10 or so line assembly routine just to generate the long numbers).

Regards,

Paul McKenzie

To expand on what Paul said, IIRC, the standard only says:

Code:

---

sizeof(char) <= sizeof(short) <= sizeof(int) <= sizeof(long)

---

So, theoretically, if a compiler implements a char as 7 bits, your longs could also be 7 bits!

Viggy

Quote:

---

Originally Posted by MrViggy

So, theoretically, if a compiler implements a char as 7 bits, your longs could also be 7 bits!

---

As I pointed out in the input unsigned long from file thread, there are minimum ranges as well. A compiler that implements a char as 7 bits would not conform to the C++ standard, since the C++ standard refers to the C standard, which mandates that CHAR_BIT be at least 8.

Quote:

---

Originally Posted by MrDoomMaster

I need access to 64-bit integrals on 32-bit architectures (Windows XP, in this example). Is there anything I can do?

---

You'll have to look at the compiler specification. Like say you use the Microsoft compiler,

http://msdn.microsoft.com/en-us/library/cc953fe1.aspx

Then you define your own long primitive somewhere globally, like

typedef __int64 LongInt;

You use LongInt at places in code where you want to make sure you have a 64 bit int. If you change compilers you just adjust the one global typedef.