Hi,
In perldoc Inline::C-Cookbook you'll find an "Object Oriented Inline" example. The following script (which works fine) is based on that example. I think the only changes I've made is to replace the strdup() calls (which are buggy) with savepv() calls - and I've added a Devel::Peek::Dump() call to reveal the innards of the objects being returned. Oh, yes ... and I've replaced malloc/free with New/Safefree:

use Devel::Peek;

my $obj1 = Soldier->new('Benjamin', 'Private', 11111);
my $obj2 = Soldier->new('Sanders', 'Colonel', 22222);
my $obj3 = Soldier->new('Matt', 'Sergeant', 33333);
    
for my $obj ($obj1, $obj2, $obj3) {
    print ($obj->get_serial, ") ", 
           $obj->get_name, " is a ", 
           $obj->get_rank, "\n");
} 

Dump($obj1);
print "\n";
 
package Soldier;
    
use Inline C => <<'END';
    
typedef struct {
  char* name;
  char* rank;
  long  serial;
  } Soldier;
 

   
SV* new(char* class, char* name, char* rank, long serial) {
    Soldier* soldier;
    New(42, soldier, 1, Soldier);
    SV*      obj_ref = newSViv(0);
    SV*      obj = newSVrv(obj_ref, class);

    soldier->name = savepv(name);
    soldier->rank = savepv(rank);
    soldier->serial = serial;
        
    sv_setiv(obj, (IV)soldier);
    SvREADONLY_on(obj);
    return obj_ref;
}
    
char* get_name(SV* obj) {
      return ((Soldier*)SvIV(SvRV(obj)))->name;
}
    
char* get_rank(SV* obj) {
      return ((Soldier*)SvIV(SvRV(obj)))->rank;
}
    
long get_serial(SV* obj) {
     return ((Soldier*)SvIV(SvRV(obj)))->serial;
}

void DESTROY(SV* obj) {
     Soldier* soldier = (Soldier*)SvIV(SvRV(obj));
     Safefree(soldier->name);
     Safefree(soldier->rank);
     Safefree(soldier);
}

END

__END__
Outputs:
11111) Benjamin is a Private
22222) Sanders is a Colonel
33333) Matt is a Sergeant
SV = PVIV(0x2f5b8c) at 0x2f4940
  REFCNT = 1
  FLAGS = (PADBUSY,PADMY,ROK)
  IV = 0
  RV = 0x2f3ca4
  SV = PVMG(0x2febbc) at 0x2f3ca4
    REFCNT = 1
    FLAGS = (OBJECT,IOK,READONLY,pIOK)
    IV = 3143860
    NV = 0
    PV = 0
    STASH = 0x45fca8    "Soldier"
  PV = 0x2f3ca4 ""
  CUR = 0
  LEN = 0
[download]

Some time ago, Ken Williams suggested a more succinct rewrite of the new() function. (See http://www.mail-archive.com/inline@perl.org/msg01370.html for the post to which I refer.) In light of Ken's post I changed the new() function to this:

SV* new(char* class, char* name, char* rank, long serial) {
    Soldier* soldier;
    New(42, soldier, 1, Soldier);

    soldier->name = savepv(name);
    soldier->rank = savepv(rank);
    soldier->serial = serial;

    return sv_setref_pv(newSViv(0), class, (void *)soldier);

}
[download]

That change, however, has a small effect on the output of the above script. Instead of getting FLAGS = (OBJECT,IOK,READONLY,pIOK) as per the above Devel::Peek::Dump(), I now get FLAGS = (OBJECT,IOK,pIOK). The READONLY flag that the original version of the script explicitly turned on is missing.

Two questions:
1) Does/could it matter that the READONLY flag is missing ?
2) How do I re-instate that READONLY flag in the modified (more succinct) version of the new() function ?

I'm inclined to think that the answer to 2) is probably straightforward and that my attempts to find the answer have simply been plagued by my own idiocy. I tried the following rendition of new() but it, too, failed to turn on any readonly flags (according to Devel::Peek::Dump, anyway):

SV* new(char* class, char* name, char* rank, long serial) {
    Soldier* soldier;
    SV *ret, *temp = newSViv(0);
    New(42, soldier, 1, Soldier);

    soldier->name = savepv(name);
    soldier->rank = savepv(rank);
    soldier->serial = serial;

    ret = sv_setref_pv(temp, class, (void *)soldier);

    SvREADONLY_on(temp);
    SvREADONLY_on(ret);

    return ret;
}
[download]

FWIW, I've long been using the technique that appears in the Inline::C Cookbook ... and I'm starting to wonder whether I ought to switch to using Ken's suggested succinctification.

Update: Part of the attraction of Ken's rewrite is that it seems to lend itself more readily to typemapping (as regards the OUTPUT section of the typemap). But I'm unsure about that, too :-)

Cheers,
Rob