Shame on me, but I had not applied the bugfixes prior to issuing "make
install". The compilation went fine after this step, and "make test" issued
no errors.
Jose
On Tue, Nov 15, 2011 at 10:04 AM, Mark Williamson <mjw.mjw.name> wrote:
> K. Uebayashi wrote:
> > Dear all,
> >
> > I'm trying to build AmberTools(patched bugfix.all: Apr. 11) with intel
> >
> >
> > <SNIP>
> > compiler> icpc -c -std=c++0x -I.. -I../../freelib -I. -o parm.o
> parm.cpp
> > parm.cpp(247): error: more than one instance of function template
> > "mort::copy_if" matches the argument list:
> > function template "_OutputIterator
> > mort::copy_if(_InputIterator, _InputIterator, _OutputIterator,
> > _Predicate)"
> > function template "_OIter std::copy_if(_IIter, _IIter,
> > _OIter, _Predicate)"
> > argument types are: (mort::diheiter_t, mort::diheiter_t,
> > std::back_insert_iterator<std::vector<mort::dihe_t,
> > std::allocator<mort::dihe_t>>>, mort::prmtop::se
> > quence_Match2)
> > copy_if( pbc.dihe_begin(), pbc.dihe_end(),
> > back_inserter(pdihs), sequence_Match2(pas) );
>
>
> I think this may be because the Intel C++ compiler (icpc) is enforcing
> the C++11 standard via the "-std=c++0x" flag. This flag is not present
> when compiling with the gnu compilers (i.e. configure gnu), but, I do
> see the same error when using gnu's g++ (4.4.1) to compile with this
> same flag.
>
> Looking at $AMBERHOME/AmberTools/src/configure, it seems that only a
> build with Intel compilers has this flag enabled:
>
> if [ "$icc_version" -ge 11 ] ; then
> cxxflags="-std=c++0x $cxxflags"
> ambercxxflags="-std=c++0x $ambercxxflags"
> ldflags="-shared-intel "
> fi
>
> I started to fix the various bits of the sleap code pertaining to this
> error in an attempt to release a fix, however, I've run into issues with
> this since as soon as I quell one error, another occurs and now I've hit
> a set that I do not know how to fix.
>
> You could try removing the -std=c++0x flag from
> ./AmberTools/src/configure and regenerate your config.h, then recompile.
> However, there must have been a reason for this flag being there in the
> first place.
>
> Ensure you run and check the test cases carefully.
>
> Regards,
>
> Mark
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Nov 15 2011 - 07:30:03 PST
