Re: Problem building netcdf-1.0.0 on Mac OS 10.8 using recent compilers.

[Alexander Hansen]

On 8/27/13 1:20 PM, Alexander Barth wrote:
> On Mon, Aug 26, 2013 at 6:23 PM, Alexander Hansen
> <address@hidden <mailto:address@hidden>> wrote:
>
>     Hi.
>
>     I'm getting pretty much the same error when I use any of the
>     following compilers that are available to Fink users on OS 10.8 to
>     install the netcdf package:
>
>     Apple's clang:
>       clang --version
>     Apple LLVM version 4.2 (clang-425.0.28) (based on LLVM 3.2svn)
>
>     Apple's llvm-gcc:
>     $ llvm-gcc-4.2 --version
>     i686-apple-darwin11-llvm-gcc-__4.2 (GCC) 4.2.1 (Based on Apple Inc.
>     build 5658) (LLVM build 2336.11.00)
>
>     FSF gcc 4.7.3 or 4.8.1 (we have earlier gcc's for prior OS versions,
>     but I'd like to be able to put out a package that works on 10.8)
>
>     The failure is as follows (slightly re-ordered):
>
>     mkoctfile __netcdf__.cc -L/sw/lib -lnetcdf
>
>     __netcdf__.cc:112:15: error: ambiguous conversion for C-style cast from
>            'octave_int<unsigned long long>' to 'size_t' (aka 'unsigned
>     long')
>                start[i] = (size_t)tmp(ndims-i-1);
>                           ^~~~~~~~~~~~~~~~~~~~~~
>     /sw/include/octave-3.6.4/__octave/oct-inttypes.h:777:3: note:
>     candidate function
>        operator T (void) const { return value (); }
>        ^
>     /sw/include/octave-3.6.4/__octave/oct-inttypes.h:781:3: note:
>     candidate function
>        operator double (void) const { return double_value (); }
>        ^
>     /sw/include/octave-3.6.4/__octave/oct-inttypes.h:783:3: note:
>     candidate function
>        operator float (void) const { return float_value (); }
>        ^

I should have added the next couple of errors:

__netcdf__.cc:135:15: error: ambiguous conversion for C-style cast from
      'octave_int<unsigned long long>' to 'size_t' (aka 'unsigned long')
          count[i] = (size_t)tmp(ndims-i-1);
                     ^~~~~~~~~~~~~~~~~~~~~~
/sw/include/octave-3.6.4/octave/oct-inttypes.h:777:3: note: candidate 
function
  operator T (void) const { return value (); }
  ^
/sw/include/octave-3.6.4/octave/oct-inttypes.h:781:3: note: candidate 
function
  operator double (void) const { return double_value (); }
  ^
/sw/include/octave-3.6.4/octave/oct-inttypes.h:783:3: note: candidate 
function
  operator float (void) const { return float_value (); }
  ^
__netcdf__.cc:156:16: error: ambiguous conversion for C-style cast from
      'octave_int<long long>' to 'ptrdiff_t' (aka 'long')
          stride[i] = (ptrdiff_t)tmp(ndims-i-1);
                      ^~~~~~~~~~~~~~~~~~~~~~~~~
/sw/include/octave-3.6.4/octave/oct-inttypes.h:777:3: note: candidate 
function
  operator T (void) const { return value (); }
  ^
/sw/include/octave-3.6.4/octave/oct-inttypes.h:781:3: note: candidate 
function
  operator double (void) const { return double_value (); }
  ^
/sw/include/octave-3.6.4/octave/oct-inttypes.h:783:3: note: candidate 
function
  operator float (void) const { return float_value (); }
  ^
3 errors generated.




>     ...
>
>     Any thoughts on a workaround?
>
>
>
> Hi,
>
> Does it work if you simply use
>
> start[i] = tmp(ndims-i-1);
>
> instead of
>
> start[i] = (size_t)tmp(ndims-i-1);
>
> at line 112 of __netcdf__.cc.
>
> Cheers,
> Alex

Thanks for the reply.

Unfortunately it looks like more tweaking will be needed.  When I 
removed the (size_t) at line 112, I got the following:

__netcdf__.cc:112:15: error: conversion from 'octave_int<unsigned long 
long>' to
      'size_t' (aka 'unsigned long') is ambiguous
          start[i] = tmp(ndims-i-1);
                     ^~~~~~~~~~~~~~
/sw/include/octave-3.6.4/octave/oct-inttypes.h:777:3: note: candidate 
function
  operator T (void) const { return value (); }
  ^
/sw/include/octave-3.6.4/octave/oct-inttypes.h:781:3: note: candidate 
function
  operator double (void) const { return double_value (); }
  ^
/sw/include/octave-3.6.4/octave/oct-inttypes.h:783:3: note: candidate 
function
  operator float (void) const { return float_value (); }
  ^
__netcdf__.cc:112:13: error: assigning to 'size_t' (aka 'unsigned long') 
from
      incompatible type 'octave_int<unsigned long long>'
          start[i] = tmp(ndims-i-1);
__netcdf__.cc:135:15: error: conversion from 'octave_int<unsigned long 
long>' to
      'size_t' (aka 'unsigned long') is ambiguous
          count[i] = tmp(ndims-i-1);
                     ^~~~~~~~~~~~~~
/sw/include/octave-3.6.4/octave/oct-inttypes.h:777:3: note: candidate 
function
  operator T (void) const { return value (); }
  ^
/sw/include/octave-3.6.4/octave/oct-inttypes.h:781:3: note: candidate 
function
  operator double (void) const { return double_value (); }
  ^
/sw/include/octave-3.6.4/octave/oct-inttypes.h:783:3: note: candidate 
function
  operator float (void) const { return float_value (); }
  ^
__netcdf__.cc:135:13: error: assigning to 'size_t' (aka 'unsigned long') 
from
      incompatible type 'octave_int<unsigned long long>'
          count[i] = tmp(ndims-i-1);
                   ^ ~~~~~~~~~~~~~~
__netcdf__.cc:156:16: error: ambiguous conversion for C-style cast from
      'octave_int<long long>' to 'ptrdiff_t' (aka 'long')
          stride[i] = (ptrdiff_t)tmp(ndims-i-1);
                      ^~~~~~~~~~~~~~~~~~~~~~~~~
/sw/include/octave-3.6.4/octave/oct-inttypes.h:777:3: note: candidate 
function
  operator T (void) const { return value (); }
  ^
/sw/include/octave-3.6.4/octave/oct-inttypes.h:781:3: note: candidate 
function
  operator double (void) const { return double_value (); }
  ^
/sw/include/octave-3.6.4/octave/oct-inttypes.h:783:3: note: candidate 
function
  operator float (void) const { return float_value (); }
  ^
5 errors generated.

--
Alexander Hansen, Ph.D.
Fink User Liaison
My package updates: http://finkakh.wordpress.com/