[Insight-developers] Rounding functions in itkMacro.h

[Sean McBride]

On 5/18/09 11:42 AM, Tom Vercauteren said:

>Regarding the two following failures on mac:
>http://www.cdash.org/CDash/viewBuildError.php?buildid=335305
>http://www.cdash.org/CDash/viewBuildError.php?buildid=335314
>
>It looks like both machines are compiling for ppc64.

Sorry I missed this discussion.  It was a holiday here in Canada yesterday.

The Rogue3 build is a ppc64-only build.  The Rogue5 build is actually a
4-way universal build (ppc32, pcc64, intel32, intel64).

>I am not familiar
>with powerpc but these failures might be related to __powerpc__ not
>being defined on ppc64. Apparently, __ppc64__ should be defined in
>this case. So changing:
>  #if defined(__GNUC__) && (!defined(__APPLE__)  || !defined(__ppc__) )
>by
>  #if defined(__GNUC__) && (!defined(__ppc__))  && (!defined(__ppc64__))
>in
>  vnl_math.h
>seems like a potential fix to me.

A 'defined(__ppc__), defined(__ppc64__)' test is legit, Apple's headers
do it sometimes too.  As for "__powerpc__" I have never seen that one,
and just searched Apple's headers and it appears nowhere.  I would stay
away from it.

But what are we really testing for here?  Is this 'fast impl' Intel
only?  If so, why not test _for_ Intel, instead of not PowerPC?  Note
that Mac OS X also builds on ARM (for iPhone), though I'm not sure
anyone is using ITK there (yet).

Cheers,

-- 
____________________________________________________________
Sean McBride, B. Eng                 sean at rogue-research.com
Rogue Research                        www.rogue-research.com 
Mac Software Developer              Montréal, Québec, Canada