Re: Installing Octave on SGI Altix

[Marco O Lanzagorta - CONTRACTOR]

Hello,

Thanks for your email. I am installing octave-2.9.17, which was
the latest "stable" version I found on the web site.

I believe the error is in the line 149 of liboctave/oct-sort.cc, which 
has the following line of code:

IFLT (pivot, *p)

and IFLT is defined as:

#define IFLT(a,b)  if (compare == NULL ? ((a) < (b)) : compare ((a), 
(b)))

Apparently it does not like the "<" symbol because it is overloaded
somewhere else in the code, and looks like it cannot determine which
is the correct one.

Any ideas?

Thanks,

Marco

On Wed Jan 30 15:05:01 -0500 2008, John W. Eaton wrote:
> On 30-Jan-2008, Marco O Lanzagorta - CONTRACTOR wrote:
> 
> | 
> | Hello,
> | 
> | I am having trouble trying to install Octave on a SGI Altix using the 
> | intel 10.0.026 icc, ifort, and icpc compilers.
> | 
> | I am using the following options:
> | 
> | FC = ifort
> | FFLAGS = -O3 -mp -ip -I/usr/local/include
> | CC = icc
> | CFLAGS = -O2 -mp -ip -I/usr/local/include
> | CXX = icpc
> | CXXFLAGS = -lstdc++ -O3 -mp -ip -I/usr/local/include
> | LDFLAGS = -L/usr/local/lib
> | 
> | On execution of the following line: 
> | 
> | icpc -c  -fPIC -I. -I.. -I../liboctave -I../src -I../libcruft/misc 
> | -DHAVE_CONFIG_H -Wall -W -Wshadow -Wold-style-cast -lstdc++ -O3 -mp -ip 
> | -I/usr\
> | /local/include ./DLD-FUNCTIONS/sort.cc -o pic/sort.o
> | 
> | I get the following error:
> | 
> | ../liboctave/oct-sort.cc(149): error: more than one operator "<" matches 
> | these operands:^M
> |             function "operator<(const octave_int<uint32_t={unsigned 
> | int}> &, const octave_int<int8_t={signed char}> &)"^M
> |             function "operator<(const octave_int<int8_t={signed char}> 
> | &, const octave_int<uint32_t={unsigned int}> &)"^M
> |             function "operator<(const octave_int<uint32_t={unsigned 
> | int}> &, const octave_int<int16_t={short}> &)"^M
> |             function "operator<(const octave_int<int16_t={short}> &, 
> | const octave_int<uint32_t={unsigned int}> &)"^M
> |             function "operator<(const octave_int<uint32_t={unsigned 
> | int}> &, const octave_int<int32_t={int}> &)"^M
> | 
> | 
> | ... and many more lines like this one.
> | 
> | Looks like the intel compiler does not like the polymorphism of the
> | Octave code.
> | 
> | Without actually having to go to change the code, do you any idea as
> | to what may be the problem and how to solve it?
> 
> What version of Octave?  Is the error message trying to say thatthe
> problem is on line 149 of sort.cc?  That's a blank line in the current
> released version of Octave (3.0.0).
> 
> If you think you've found a bug, please send a complete report to
> address@hidden  Please also read http://www.octave.org/bugs.html to
> see what information you should include so someone might be able to
> help you.
> 
> jwe

-- 
===========================================================================
Dr. Marco Lanzagorta (Contractor)     email: address@hidden
Research Computers Group - Code 5593  phone: (202) 767-8427
Center for Computational Science      fax  : (202) 404-7402
US Naval Research Laboratory
4555 Overlook Ave, S.W.
Washington, DC 20375
===========================================================================


----- End forwarded message -----

-- 
===========================================================================
Dr. Marco Lanzagorta (Contractor)     email: address@hidden
Research Computers Group - Code 5593  phone: (202) 767-8427
Center for Computational Science      fax  : (202) 404-7402
US Naval Research Laboratory
4555 Overlook Ave, S.W.
Washington, DC 20375
===========================================================================