«

Hi,

If I'm doing something like this:

#include<stdint.h>

uint64_t i = (0xffff * 0xffff ) ;

Or indeed any arbitrarily long expression, how do I ensure the
intermediate results won't be truncated to 32 bits without putting
irritating (uint64_t) casts next to each operand? And how do I get a
default sizeof(int) == 8 on an IA-64/G5?

Richard Cavell wrote:
> Hi,
>
> If I'm doing something like this:
>
> #include<stdint.h>
>
> uint64_t i = (0xffff * 0xffff ) ;

Wrong group, ask in comp.lang.c
( This is C99, in comp.lang.c++ only C++98/C++2003 are on-topic
and they are not based on C99 )

Regards,
Michiel Salters

On 6/4/05 7:31 PM, msalters wrote:
> Richard Cavell wrote:
>
>>Hi,
>>
>>If I'm doing something like this:
>>
>>#include<stdint.h>
>>
>>uint64_t i = (0xffff * 0xffff ) ;
>
>
> Wrong group, ask in comp.lang.c
> ( This is C99, in comp.lang.c++ only C++98/C++2003 are on-topic
> and they are not based on C99 )

Okay, then what's the equivalent C++ way to do it?

Richard Cavell wrote:

> Hi,
>
> If I'm doing something like this:
>
> #include<stdint.h>
>
> uint64_t i = (0xffff * 0xffff ) ;
>
> Or indeed any arbitrarily long expression, how do I ensure the
> intermediate results won't be truncated to 32 bits without putting
> irritating (uint64_t) casts next to each operand? And how do I get a
> default sizeof(int) == 8 on an IA-64/G5?


If your question is about C, you should ask in comp.lang.c.


If your question is .NET related (Visual C++ .NET) you should ask in
microsoft.public.dotnet.languages.vc newsgroup.

If it doesn't appear in your news server use the public MS news server:

msnews.microsoft.com




--
Ioannis Vranos

http://www23.brinkster.com/noicys

Richard Cavell wrote:
> On 6/4/05 7:31 PM, msalters wrote:
> > Richard Cavell wrote:
> >
> >>Hi,
> >>
> >>If I'm doing something like this:
> >>
> >>#include<stdint.h>
> >>
> >>uint64_t i = (0xffff * 0xffff ) ;
> >
> >
> > Wrong group, ask in comp.lang.c
> > ( This is C99, in comp.lang.c++ only C++98/C++2003 are on-topic
> > and they are not based on C99 )
>
> Okay, then what's the equivalent C++ way to do it?

unsigned long i = (0xffffUL*0xffffUL);

which will not overflow, but I don't see what 64 bits code has to
do with it. Besides, IIRC math base 2^32 doesn't care about overflows
on multiplications. ~0UL*~0UL==01UL, even if I use a 64-bits
intermediate.