RE: PP Library...

From: Eric Robert (synapzzz_at_[hidden])
Date: 2003-05-06 13:19:59

Ok,
I see that there are 3 ways of doing this now. What is the
recommendation for something small like 2 or 3 lines for each native
types i.e. more or less 12 iterations?

Eric

"Paul Mensonides" <yg-boost-users_at_[hidden]> wrote
> Tanton Gibbs wrote:
> > You can use BOOST_PP_LIST_FOR_EACH
> >
> > #define ARG_LIST (char, (int, (unsigned int, (long, ...))))
> > #define DO_IT(r, data, elem) void foo( elem );
> >
> > BOOST_PP_LIST_FOR_EACH( DO_IT, , ARG_LIST )
>                                 ^
> This empty argument is undefined in C++.  In C99, it is allowed and
called
a
> "placemarker."  Instead of passing nothing, just pass anything and
ignore
it.
>
> // ----- //
>
> #include <boost/preprocessor/list/for_each.hpp>
>
> #define ARG_LIST \
>     (char, (int, (unsigned, (long, BOOST_PP_NIL)))) \
>     /**/
> #define DO_IT(r, ignored, elem) \
>     void foo( elem ); \
>     /**/
>
> BOOST_PP_LIST_FOR_EACH( DO_IT, ?, ARG_LIST )
>
> #undef ARG_LIST
> #undef DO_IT
>
> // ----- //
>
> Note that there are other ways to do this also.  For instance, use of
> "sequences" makes the definition of ARG_LIST easier:
>
> // ----- //
>
> #include <boost/preprocessor/seq/for_each.hpp>
>
> #define ARGS \
>     (char)(int)(unsigned)(long) \
>     /**/
> #define DO_IT(r, ignored, elem) \
>     void foo( elem ); \
>     /**/
>
> BOOST_PP_SEQ_FOR_EACH( DO_IT, ?, ARGS )
>
> #undef ARGS
> #undef DO_IT
>
> // ----- //
>
> Of course, if what you want to generate is much larger, you might want
to
use
> some form of vertical repetition:
>
> // ----- //
>
> #include <boost/preprocessor/iteration/local.hpp>
> #include <boost/preprocessor/seq/elem.hpp>
>
> #define ARGS \
>     (char)(int)(unsigned)(long) \
>     /**/
>
> #define BOOST_PP_LOCAL_MACRO(n) \
>     void foo( BOOST_PP_SEQ_ELEM(n, ARGS) ); \
>     /**/
> #define BOOST_PP_LOCAL_LIMITS \
>     (0, BOOST_PP_SEQ_SIZE(ARGS) - 1) \
>     /**/
> #include BOOST_PP_LOCAL_ITERATE()
>
> #undef ARGS
>
> // ----- //
>
> Regards,
> Paul Mensonides

 

[Non-text portions of this message have been removed]