RE: [boost] Re: BOOST_PP_SEQ_ENUM

From: Paul Mensonides (pmenso57_at_[hidden])
Date: 2005-01-25 16:06:04

> -----Original Message-----
> From: boost-bounces_at_[hidden] 
> [mailto:boost-bounces_at_[hidden]] On Behalf Of Arkadiy Vertleyb

> > BOOST_PP_SEQ_ENUM_3[space](x)(y)(z)
> >
> > and stops here, probably because of the space.
> 
> Please disregard this -- it's definitely not a space problem. 
>  There appears to be something wrong with my usage of it -- 
> sometimes it works, and sometimes it doesn't.  Something like 
> this doesn't work as I would expect:
> 
> cout << BOOST_PP_STRINGIZE(BOOST_PP_SEQ_ENUM((x)(y)(z)));

That shouldn't work regardless.  An invocation of SEQ_ENUM with anything other
than one element becomes (what I call) an intermediate.  An "intermediate" is an
argument to a macro that expands to multiple arguments.  There is no way that
STRINGIZE could be defined to handle an intermediate.  E.g.

#define STRINGIZE(x) PRIMITIVE_STRINGIZE(x)
#define PRIMITIVE_STRINGIZE(x) #x

The delay is necessary to allow 'x' to expand on input to STRINGIZE.  However,
because it does expand in this case, it tries to invoke:

PRIMITIVE_STRINGIZE(x, y, z)

...which is, of course, too many arguments.

> But in other contexts it seems to work fine.
> 
> Also, if I replace BOOST_PP_SEQ_ENUM with the following:
> 
> #define BOOST_PP_SEQ_ENUM(seq) 
> BOOST_PP_CAT(BOOST_PP_CAT(BOOST_PP_SEQ_ENUM_,
> BOOST_PP_SEQ_SIZE(seq)), seq)
> 
> The output is "x, y, z" as  expected.

Did you try the other workaround that I sent?  The problem with the above is
that it is picking up expansions that should have happened already (but haven't
due to VC's bug-ridden implementation).  The problem is that the above creates a
dependency between CAT and whatever expansion might be picked up.

Regards,
Paul Mensonides