3.0.2 build failure on RHELv5.1

[Jeff Blaine]

John W. Eaton wrote:
> On 22-Sep-2008, Jeff Blaine wrote:
> 
> | John W. Eaton wrote:
> | > On 22-Sep-2008, Jeff Blaine wrote:
> | > 
> | > | The zlib version that comes with RHEL5.1, which appears to be
> | > | 1.2.3
> | > | 
> | > | octave-3.0.2:beijing> rpm -qa | grep zlib
> | > | zlib-1.2.3-3
> | > | zlib-1.2.3-3
> | > | zlib-devel-1.2.3-3
> | > | zlib-devel-1.2.3-3
> | > | octave-3.0.2:beijing>
> | > 
> | > Does it have the gzclearerr?  Does the header file provided with this
> | > package declare gzclearerr?
> | > 
> | > Looking at Octave's configure.in script, it checks for the presence of
> | > gzclearerr when checking for zlib.  So I don't understand why that
> | > check would have succeeded if gzclearerr is not present.
> | > 
> | > jwe
> | 
> | John,
> | 
> | octave-3.0.2:beijing> grep gzclearerr /usr/include/zlib.h
> | ZEXTERN void ZEXPORT gzclearerr OF((gzFile file));
> | octave-3.0.2:beijing>
> | 
> | octave-3.0.2:beijing> grep zlib myconfigure.log
> | checking zlib.h usability... yes
> | checking zlib.h presence... yes
> | checking for zlib.h... yes
> | octave-3.0.2:beijing>
> 
> So from this, I assume HAVE_ZLIB_H is defined in config.h, correct?
> 
> If so, then I don't see the cause for the error.  Is it possible there
> is some other zlib.h that is being included instead of the one in
> /usr/include?
> 
> jwe
> 

That's correct:

octave-3.0.2:beijing> grep ZLIB config.h
/* Define if ZLIB is available. */
#define HAVE_ZLIB 1
#define HAVE_ZLIB_H 1
octave-3.0.2:beijing>

It's possible there is some other zlib.h being included, but
there is no reasonable explanation as to why.  For example:

octave-3.0.2:beijing> grep gzclearerr /usr/rcf/include/zlib.h
octave-3.0.2:beijing>

However, I have nothing whatsoever in my shell's environment
that suggests /usr/rcf anywhere.

octave-3.0.2:beijing> env | grep /usr/rcf
octave-3.0.2:beijing>