This is the mail archive of the gcc-help@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

C preprocessor For loops ?



#include <stdio.h>


/* explore ForX() macro expansion possibilities, mainly for use in
  array initialization expressions

Want ForX(N, Do_Macro(X))
- where param X in Do_Macro goes from 0..N (N is the value passed in)
- domacro initializes fields in elements consistent with their position in the list.


  Breakdown:
  - create some fixed-N expanders,
  - add them up as needed.
  - each N-exander gets its starting-point

  Assumptions
  - ForX(N,Func) presumably is buildable for arbitrary N
  - it may have to emit the "struct a X[] = { }" part.

  Problems And Question areas
  - X <- N
  - scanning and rescanning
  - influence of #, ## on rescans
*/



#define Given(N, C99Expr...)    \
   [N] = { .index = N, C99Expr },

#define FOR_1(N, Rest...)    \
   Given(N, Rest)

#define FOR_2(N, Rest...)    \
   FOR_1(N, Rest)        \
   FOR_1(1+N, Rest)    \

#define FOR_4(N, Rest...)    \
   FOR_2(N, Rest)        \
   FOR_2(2+N, Rest)    \

#define FOR_8(N, Rest...)    \
   FOR_4(N, Rest)        \
   FOR_4(4+N, Rest)

#define FOR_16(N, Rest...)    \
   FOR_8(N, Rest)        \
   FOR_8(8+N, Rest)

#define FOR_32(N, Rest...)    \
   FOR_16(N, Rest)        \
   FOR_16(16+N, Rest)

/* The above works, but has at least 1 usage caveat:

  Commas are added only by Given(), any others added - either in the
  FOR_N() macros, or in the final use, will create compiler errors.

  Its easy enough to use correctly, but compler errors are cryptic
  when you dont:
   foo.c:49: error: syntax error before ',' token
*/

struct a {
   int index;
   int i,j;
   int foo;
};

// no terminating comma here
#define COMMON_PARMS(N)  .i=(N)+1, .j=2*(N)

// try wrapping mess
#define MAKE_ARR(NAME, N, ELEM_GEN)    struct a NAME[] = { ELEM_GEN(N) }
// didnt help
// MAKE_ARR ( X, 3, (COMMON_PARMS(N), .foo=10) );
// MAKE_ARR ( X, 4, FOR_4(N, COMMON_PARMS(0), .foo=10) );

MAKE_ARR ( X, 4, FOR_4(N, COMMON_PARMS(N), .foo=10) );

struct a Y[] = {
   FOR_2(0, COMMON_PARMS(0), .foo=10)
   FOR_2(2, COMMON_PARMS(1), .foo=11)
};

int main(int c, char **v)
{
   int i;
   for (i=0; i<sizeof(X)/sizeof(struct a); i++) {
       printf("%d x:%d i:%d j:%d\n", i, X[i].index, X[i].i, X[i].j);
   }
}

/* when it runs, it does:

0 x:0 i:1 j:0
1 x:1 i:1 j:0
2 x:2 i:2 j:2
3 x:3 i:2 j:2

Things partly work:

- x is getting value of N, properly propagated thru the 2 uses of the
 FOR-2 macro

- i,j are not following N directly, their N comes from COMMON_PARMS(),
so its value is bound, when 1st expanded.


Why Cant it Work ?


I suspect what Im after cant be done, and Ill guess why - pls comment here.

- Im trying to late-bind N, in-side the For-N macros

- in lisp-ese: (forgive me)

FOR_2(2, 'COMMON_PARMS(N), .foo=11)

- I cant use N in the X-decl, since its not a macro invocation, and N
 doesnt exist.  TRY THIS. nope

- is there something with # or ## that could force a rescan, or does this quote preclude it?

6.10.3.1 Argument substitution
1 After the arguments for the invocation of a function-like macro have been identified,
argument substitution takes place. A parameter in the replacement list, unless preceded
by a # or ## preprocessing token or followed by a ## preprocessing token (see below), is
replaced by the corresponding argument after all macros contained therein have been
expanded. Before being substituted, each argument's preprocessing tokens are
completely macro replaced as if they formed the rest of the preprocessing file; no other
preprocessing tokens are available.



Ive run out of ideas here, any help ?


*/


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]