I compiled LAMMPS using intel compiler and I got a warning says
"
…/pair_list.cpp(52): warning #47: incompatible redefinition of macro “_noalias” (declared at line 196 of “…/lmptype.h”) #define _noalias __restrict
^
"
the pair_list lines are ( from 193 to 201 ):
"
// declaration to lift aliasing restrictions
#if defined(__INTEL_COMPILER) #define _noalias restrict ( this is the line 196 ) #elif defined(GNUC) #define _noalias __restrict #else #define _noalias #endif
"
is there any mistake in this line, like restrict must have a “__” before it ?
I compiled LAMMPS using intel compiler and I got a warning says
"
../pair_list.cpp(52): warning #47: incompatible redefinition of macro
"_noalias" (declared at line 196 of "../lmptype.h") #define _noalias __restrict
^
"
the pair_list lines are ( from 193 to 201 ):
"
// declaration to lift aliasing restrictions
#if defined(__INTEL_COMPILER) #define _noalias restrict ( this is the line 196 ) #elif defined(__GNUC__) #define _noalias __restrict #else #define _noalias #endif
"
is there any mistake in this line, like restrict must have a "__" before it
no. the mistake is in your compiler flags.
as has been explained *many* times on this mailing list: when using
the intel compiler you *must* use the -restrict flag.
Mr. Kohlmeyer i am sorry i found the reason of this warning ( noalise warning ).
the reason is that i added an external package to lammps which is written to fit the 10Aug15 version of lammps and this package edits the file pair_list and return it to the old plus and it contains these lines
"
../manifold_gaussian_bump.cpp(371): warning #268: the format string ends
before this argument
gg, nn[0], nn[1], nn[2] );
^
"
still appearing.
unless you are planning to do development in the USER-MANIFOLD
package, you should not worry about it.
it is a bug, but not a critical one. it only affects function
generating test output for developers.
>
> but the second warning which is
>
> "
> ../manifold_gaussian_bump.cpp(371): warning #268: the format string
ends
> before this argument
> gg, nn[0], nn[1], nn[2] );
> ^
> "
>
> still appearing.
unless you are planning to do development in the USER-MANIFOLD
package, you should not worry about it.
it is a bug, but not a critical one. it only affects function
generating test output for developers.
axel.
Indeed. FYI, it is fixed in later versions but it is pretty much harmless.
If you are in fact using USER-MANIFOLD you might want to update to the
latest version because there were some issues with the Gaussian bump
manifold that got fixed in later versions. Otherwise there's no real harm.
>
> but the second warning which is
>
> "
> ../manifold_gaussian_bump.cpp(371): warning #268: the format string
> ends
> before this argument
> gg, nn[0], nn[1], nn[2] );
> ^
> "
>
> still appearing.
unless you are planning to do development in the USER-MANIFOLD
package, you should not worry about it.
it is a bug, but not a critical one. it only affects function
generating test output for developers.
axel.
Indeed. FYI, it is fixed in later versions but it is pretty much harmless.
Me and my big mouth. I apparently fixed it locally but never propagated it upstream. The function was never called anywhere anyway so it’s mostly a change to get rid of a compiler warning, but still, thanks for cleaning up.