Re: Octave 2.9.16 available for ftp -- build problems on freebsd/amd64

[sam sirlin]

John W. Eaton wrote:
> On 14-Nov-2007, sam sirlin wrote:
>
> | Regarding Makefrag.bsd, the flags that are filtered out for some files
> | 
> |   PT_FLAGS = -fexternal-templates -fno-implicit-templates
> |   CXXFLAGS_NO_PT_FLAGS = $(filter-out $(PT_FLAGS), $(ALL_CXXFLAGS))
> | 
> | are never set in my normal compilation. In fact, some tests seem to show 
> | that  the compile lines with
> | 
> | CXXFLAGS_NO_PT_FLAGS
> | 
> | are never used anyway. Not sure why.
> | rm src/pic/lex.o; gmake -d gives tons of output, including
> |          Found an implicit rule for `pic/lex.o'.
> | compilation line without filter
> | 
> | adding to cxxflags I do see
> | c1plus: warning: switch '-fexternal-templates' is no longer supported
> | 
> | I'll try and do a complete build from scratch forcing those flags in 
> | cxxflags later...
>
> OK, I looked again and given that Octave's configure script no longer
> tries to use these options (and probably hasn't in several years at
> least) I think it should be safe to remove Makefrag.bsd.
>   

To check, I ran with
CXXFLAGS = -fPIC -fexternal-templates -fno-implicit-templates

and disabled the filter, and it did not build; stopped with a whole slew 
of undefined referrences

g++42 -I/usr/local/include -I/usr/local/include/UMFPACK 
-I/usr/local/include/metis -I. -I.. -I../liboctave -I../src 
-I../libcruft/misc   -DHAVE_CONFIG_H  -Wall -W -Wshadow -Wold-style-cast 
-fPIC -fexternal-templates -fno-implicit-templates -rdynamic \
   -L..  -fPIC -L/usr/local/lib -lmetis -lm  -o octave \
   main.o  \
   -L../liboctave -L../libcruft -L../src -Wl,-rpath 
-Wl,/usr/local/lib/octave-2.9.16 \
   -loctinterp -loctave  -lcruft    \
    -lumfpack -lamd -lcamd -lcolamd \
   -lcholmod -lccolamd -lcxsparse -llapack -lcblas -lf77blas -latlas \
   -lfftw3 -lreadline  -lncurses -lz -lm  -L/usr/local/lib 
-L/usr/local/lib/gcc-4.2.3/gcc/x86_64-portbld-freebsd6.2/4.2.3 
-L/usr/local/lib/gcc-4.2.3/gcc/x86_64-portbld-freebsd6.2/4.2.3/../../.. 
-lmetis -lgfortranbegin -lgfortran -lm -lz
../liboctave/liboctave.so: warning: warning: tempnam() possibly used 
unsafely; consider using mkstemp()
../liboctave/liboctave.so: undefined reference to `bool 
operator!=<unsigned long, signed char>(octave_int<unsigned long> const&, 
octave_int<signed char> const&)'
../src/liboctinterp.so: undefined reference to 
`std::deque<std::basic_string<char, std::char_traits<char>, 
std::allocator<char> >, std::allocator<std::basic_string<char, 
std::char_traits<char>, std::allocator<char> > > 
>::_M_destroy_data_aux(std::_Deque_iterator<std::basic_string<char, 
std::char_traits<char>, std::allocator<char> >, std::basic_string<char, 
std::char_traits<char>, std::allocator<char> >&, std::basic_string<char, 
std::char_traits<char>, std::allocator<char> >*>, 
std::_Deque_iterator<std::basic_string<char, std::char_traits<char>, 
std::allocator<char> >, std::basic_string<char, std::char_traits<char>, 
std::allocator<char> >&, std::basic_string<char, std::char_traits<char>, 
std::allocator<char> >*>)'
..

so I guess if you ever bring back the switches the filter might be 
necessary... but if not then the filter of course isn't needed.