64 bit integer

[Jay]

Hello, I am sure this has a quick and easy solution but I can't find
it.

I need to store 19 '9's in an integer which means i need an unsigned 64
bit
integer.
in Visual Studio the code would be:

__int64 test = 9999999999999999999;

I am using g++ version 3.3.5(suse 9.3) and have tried both:

unsigned long long test = 999999999999999999;
and:
uint64_t test = 9999999999999999999;

both have this compile error using the command "g++ test.cpp"
test.cpp:5: error: integer constant is too large for "long" type

Am I forgeting to include something or not passing an option to g++, or
is
there another way to do 64 bit integer storage?

Thank you for your help

 

[Michiel.Salters]

Hello, I am sure this has a quick and easy solution but I can't find
it.

I need to store 19 '9's in an integer which means i need an unsigned 64
bit integer. in Visual Studio the code would be:

__int64 test = 9999999999999999999;

The double underscore is a giveaway that this isn't C++, but a
Microsoft
extension.

I am using g++ version 3.3.5(suse 9.3) and have tried both:

unsigned long long test = 999999999999999999;
and:
uint64_t test = 9999999999999999999;

both have this compile error using the command "g++ test.cpp"
test.cpp:5: error: integer constant is too large for "long" type

Reasonable, by the standard rules it may indeed reject the code. If
you need an GCC extension to C++, you should ask in a GCC group.

HTH,
Michiel Salters

 

[mlimber]

Hello, I am sure this has a quick and easy solution but I can't find
it.

I need to store 19 '9's in an integer which means i need an unsigned 64
bit
integer.

[snip]

Or you could use a number class with arbitrary precision. The sizes of
the various built-in integral types are platform-dependent, so you'll
want to check with your OS/compiler documentation.

Cheers! --M

 

[Jay]

So is there no way to do this in C++ without using an extension?
If this is the wrong group for this question, do you happen to know
which group would be more appropriate?

Thank you

 

[Jay]

I know that the sizes of int, long etc are platform-dependent and I
know that they are not big enough on my platform, I am looking for a
datatype that guarentees me 64bits on any platform. I don't really want
to have to include non-standard libraries, but is there a number class
with arbitrary precision that you would suggest?

 

[Bronek Kozicki]

uint64_t test = 9999999999999999999;

both have this compile error using the command "g++ test.cpp"
test.cpp:5: error: integer constant is too large for "long" type

try:

uint64_t test = 9999999999999999999LL;


B.

 

[mlimber]

I know that the sizes of int, long etc are platform-dependent and I
know that they are not big enough on my platform, I am looking for a
datatype that guarentees me 64bits on any platform.

No such thing with built-in types. C++ compilers can be standard
compliant on 8-bit machines, which implies that no built-in type
*necessarily* supports 64 bits.

I don't really want
to have to include non-standard libraries, but is there a number class
with arbitrary precision that you would suggest?

Check out:

http://www.oonumerics.org/oon/

Cheers! --M

 

[mlimber]

Check out:

http://www.oonumerics.org/oon/

See also

http://www.swox.com/gmp/

Cheers! --M

 

[James.Mahler]

unsigned long long a = 9999999999999999999LLU; is the right way to do
it. If you don't say it's unsigned you get warnings because it's too
large.

That works fine for 64 bit numbers on 32 bit machine.

 

[Jay]

Thanks, I just forgot the LLU on the end of a non-int constant

 

[mlimber]

unsigned long long a = 9999999999999999999LLU; is the right way to do
it. If you don't say it's unsigned you get warnings because it's too
large.

That works fine for 64 bit numbers on 32 bit machine.

long long is a compiler extension. It is not standard and will not work
on all platforms or compilers. long is guaranteed to be at least 32
bits by the standard; that's as big as it gets as far as the standard
is concerned (for now).

Cheers! --M

 

[Clark S. Cox III]

unsigned long long a = 9999999999999999999LLU; is the right way to do
it.

That is, if your C++ compiler supports 'long long' as an extension.
While many do, remember that 'long long' is a feature of C99, *not* of
C++.

 

[Greg]

Hello, I am sure this has a quick and easy solution but I can't find
it.

I need to store 19 '9's in an integer which means i need an unsigned 64
bit
integer.
in Visual Studio the code would be:

__int64 test = 9999999999999999999;

I am using g++ version 3.3.5(suse 9.3) and have tried both:

unsigned long long test = 999999999999999999;
and:
uint64_t test = 9999999999999999999;

both have this compile error using the command "g++ test.cpp"
test.cpp:5: error: integer constant is too large for "long" type

Am I forgeting to include something or not passing an option to g++, or
is
there another way to do 64 bit integer storage?

Yes. This is how the declaration should look:

unsigned long long test = 999999999999999999ULL;

long long constants require an LL or ULL type designator. Although long
longs are not yet officially part of the C++ language they will be
added in the next revision.

Greg

 

[Greg]

long long is a compiler extension. It is not standard and will not work
on all platforms or compilers. long is guaranteed to be at least 32
bits by the standard; that's as big as it gets as far as the standard
is concerned (for now).

long longs are part of the C99 standard and will be in the next
revision of the C++ standard.

It is C99, and not C++, that specifies the size of integer types. C++
makes no guarantee that any integer type will be larger than a byte.

Greg

 

[Jay]

Well I think I will go with the extension that doesn't require me to go
find another class to add to my code which should be simple for a
programing language as powerful as c++. I am disappointed that
everything I want to do for this project is an extension of c++. Maybe
I
should just report that the project is coded in a custom language. At
least it's working with long long, hopefully whatever commitee decides
on c++ standards will help me out someday, but until then, I'm going to
be a fan of extensions

 

[Jay]

Good to hear, I'm guessing that if it's known to be part of the next
revision, the compilers will have already put it in. So nothng to
really worry about there

 

[Markus Moll]

Hi

Jay wrote:

[about long long]

Good to hear, I'm guessing that if it's known to be part of the next
revision, the compilers will have already put it in. So nothng to
really worry about there

But following the reasoning of your recent post elsewhere in this thread,
you would soon start crying that there is no type long long long that is
guaranteed to be at least 96 or 128 bits long. And if one would add this,
at some point the need for an 256 bits type would come up. A never-ending
story...

If you're targeting a specific platform, you're fine using whatever your
compiler offers you (i.e. extensions).

If you want a program that is standard compliant and thus
platform-independent, you will have to use some sort of arbitrary precision
library. (or at least a library that defines some sort of 64-bit integer (I
think boost does?))

If you're about to argue that this should be part of the standard library:
I'm with you there.

But please don't blame the language.

cheers
Markus

 

[Jay]

first about your story. I hope you are not naive enough to think
anything in software should have an ending.
The fact that you are "with me" on the argument that it should be in
the standard shows that you agree that
the language should change as times change.

Since you brought reasoning how do you logically say that the language
should have something it doesn't, and don't blame the language?


The reason I wanted, and apparently was crying over, a 64 bit int
only has a little to do with being a baby.
I am just trying to grab a value from a database, a value that is a 64
bit number. it's one variable in my program that I get and then do a
couple
computations on. Forgive me if I didn't want to use your precious
boost library and add an extra file to my project for 1 variable. The
point is that
the database is a pretty old database and I figured that if it was easy
for the database people to do years ago it should be cake today.
Adding a library is less
than cake. long long is a little closer to cake. It being a standard
and documented would be cake. I just want cake or as close to it as I
can get.

That being said I don't know why you posted. I thought "Good to hear"
was a positive notion.

 

[Jack Klein]

unsigned long long a = 9999999999999999999LLU; is the right way to do
it. If you don't say it's unsigned you get warnings because it's too
large.

That works fine for 64 bit numbers on 32 bit machine.

Except that there is no long long type, signed or unsigned, in
standard C++. It is standard in C since 1999, but an extension in
C++.

 

[Bronek Kozicki]

the database is a pretty old database and I figured that if it was
easy for the database people to do years ago it should be cake today.

hahahaha. Tell me more about fixed point decimal arithmetic in C++ .


B.