Re: [boost] [preprocessor] nested for_each not working like expected

Subject: Re: [boost] [preprocessor] nested for_each not working like expected
From: Nathan Ridge (zeratul976_at_[hidden])
Date: 2013-08-07 16:01:41

> >> I can't figure out what I'm doing wrong. According to the docs, this
> >> should work:
> >>
> >> #define S0 (0)(1)(2)(3)
> >> #define S1 (5)(6)(7)(8)
> >> #define M4(R, DATA, ELEM) (DATA,ELEM)
> >> #define M2(R, DATA, ELEM) BOOST_PP_SEQ_FOR_EACH_R(R, M4, ELEM, S1);
> >> BOOST_PP_SEQ_FOR_EACH(M2, ~, S0)
> >>
> >> I expect to get a sequence like (0,1)(0,2)... for each combination. I
> >> get gibberish. (Please don't tell me to use FOR_EACH_PRODUCT. In my real
> >> use case, M2 and M4 are far more complicated, and a simple product won't
> >> do.)
> >>
> >> Can anybody please tell me what I'm doing wrong? TIA,
> > 
> > According to [1], only BOOST_PP_FOR, BOOST_PP_WHILE, and BOOST_PP_REPEAT
> > are reentrant. Other looping macros, like SEQ_FOR_EACH, are not.
> 
> Yes, but [2] indicates that SEQ_FOR_EACH is implemented in terms of FOR
> and has a _R version for reentry. Why is it provided if it doesn't work?
> 
> [2]
> http://www.boost.org/doc/libs/1_54_0/libs/preprocessor/doc/ref/seq_for_each_r.html

My understanding is that SEQ_FOR_EACH is reentrant with FOR, but not with itself.
So one of the inner and outer loops can use SEQ_FOR_EACH, and the other FOR, and
this works fine despite the fact that SEQ_FOR_EACH is implemented in terms of for.
However they cannot both use SEQ_FOR_EACH.

Also, I believe the _R versions are for efficiency only - that, is, the make re-entrancy
more efficient in cases where it would already be possible, just less efficient, without
the _R.

I'm not certain about this, though. Perhaps Paul or Edward can clarify these points.

Regards,
Nate