[Insight-users] [Vxl-users] fail to compile on OpenBSD

Amit Kulkarni amitkulz at gmail.com
Fri Oct 8 18:50:35 EDT 2010


Hi Ian,

Sorry to bother you but Brad King @ ITK-users pointed out a cleaner
way of resolving the patch for swap32/swap64 on OpenBSD. Could you
back out the earlier changes and use this one? I compiled it and it
was fine.

http://review.source.kitware.com/#patch,sidebyside,106,2,Utilities/vxl/core/vnl/vnl_matlab_header.h

Also, please ignore this patch from previous email, Brad points out
that these are macros. I don't know what I was doing, just trying out
a new compiler pcc.

So the only things to get Vxl to compile is your vnl_is_nan() patch
and this undef swap32/swap64 patch

Thanks

On Wed, Oct 6, 2010 at 10:13 PM, Amit Kulkarni <amitkulz at gmail.com> wrote:
> Hi Ian,
>
> I found out small minor stuff using pcc as c compiler (installed in
> /usr/local/bin/pcc) which is probably applicable to vxl. This diff is
> against itk's trunk version, pardon me for not using vxl's trunk but
> it will be almost the same result.
>
> In $VXL-source/Utilities/vxl/v3p/netlib/lapack/complex16/zlatrs.h
> there is a spurious bracket '(' inside that extern int. I believe on
> line 8. I don't know why this is not showing up on git diff
>
> ------------------------ additional git diff's follows-----------------------
>
> diff --git a/Utilities/vxl/vcl/vcl_cmath.h b/Utilities/vxl/vcl/vcl_cmath.h
> index 7a09c86..0ce83b9 100644
> --- a/Utilities/vxl/vcl/vcl_cmath.h
> +++ b/Utilities/vxl/vcl/vcl_cmath.h
> @@ -61,7 +61,7 @@
>  #endif
>
>
> -#if !VCL_COMPLEX_POW_WORKS && !defined VCL_CMATH_POW_DECLARED
> +#if !defined(VCL_COMPLEX_POW_WORKS) && !defined (VCL_CMATH_POW_DECLARED)
>  // deal with corrections to pow(complex...)
>  # undef vcl_pow
>  # define vcl_pow vcl_pow
> diff --git a/Utilities/vxl/vcl/vcl_complex.h b/Utilities/vxl/vcl/vcl_complex.h
> index 93cc8ff..12e2aaa 100644
> --- a/Utilities/vxl/vcl/vcl_complex.h
> +++ b/Utilities/vxl/vcl/vcl_complex.h
> @@ -63,7 +63,7 @@
>  #endif
>
>
> -# if !VCL_COMPLEX_POW_WORKS && !defined(VCL_SGI_CC_7)
> +# if !defined(VCL_COMPLEX_POW_WORKS) && !defined(VCL_SGI_CC_7)
>  #  undef vcl_pow
>  #  define vcl_pow vcl_pow
>  // several implementations of pow are wrong.
> diff --git a/Utilities/vxl/vcl/vcl_valarray.h b/Utilities/vxl/vcl/vcl_valarray.h
> index e024aee..16bf761 100644
> --- a/Utilities/vxl/vcl/vcl_valarray.h
> +++ b/Utilities/vxl/vcl/vcl_valarray.h
> @@ -20,7 +20,7 @@
>  # include "iso/vcl_valarray.h"
>  #endif
>
> -#if !VCL_COMPLEX_POW_WORKS
> +#if !defined(VCL_COMPLEX_POW_WORKS)
>  // deal with corrections to pow(complex...)
>  # undef vcl_pow
>  # define vcl_pow vcl_pow
>
> Thanks
>
> On Wed, Oct 6, 2010 at 3:25 AM, Ian Scott <scottim at imorphics.com> wrote:
>> vnl_math and vnl_mathlab_* fixes committed to the trunk.
>>
>> Thanks.
>> Ian.
>


More information about the Insight-users mailing list