Re: determining 32 v. 64 bit compilatio

[Andrew W. Nosenko]

On Wed, Jun 30, 2010 at 11:55, Wesley Smith <address@hidden> wrote:
>>> I don't disagree, but let's consider this an academic exercise at this
>>> point.  I'd actually like to know how to do this regardless of whether
>>> it violates anyones code sensibilities.
>>
>> Before we answer your question, I'd like a clarification: what would be the 
>> proper setting on OS X for you, when the compiler generates both 32-bit and 
>> 64-bit code in the same object?
>>
>> In other words, are you looking for a configuration variable that changes 
>> the way your code is compiled, or are you looking for a way to change the 
>> name of the files that are generated by the compiler?
>
> On OSX, I'm looking for something that adapts to the -arch i386 -arch
> x86_64 flags, so it should depend on the compiler settings, not
> generate them.  Looking further into the library (which is open source
> fortunately), I find where the flag is used:
>
>
> #ifdef x86_64
>
> typedef signed int int32;
> typedef unsigned int uint32;
>
> #else
>
> typedef signed long int32;
> typedef unsigned long uint32;
>
> #endif
>
>
> In the end it looks like I need to set this based on the size of int
> and long, so calls like AC_CHECK_SIZEOF(unsigned long) seem to be most
> appropriate.  Thanks for all of the help.  This is my first time doing
> anything with autoconf.  It's incredibly helpful to get some much good
> info.  Sorry if my questions seem vague.  I'm only just starting to
> get a hold of the issues involved in writing a portable build system.
>

Excuse me, but why just don't

    #include <inttypes.h>

and use int32_t and uint32_t declared there?

If you want to deal with some heavy non-posix, or just too old system, then

    AC_TYPE_INT32_T

autoconf macro (and like) may help you.

-- 
Andrew W. Nosenko <address@hidden>