Hmmm. Now you understand why I said "called but not used the comparator function" back up there somewhere. I could never get the values I was stacking to show up in the sub either?

I simplified things by moving all the GV stuff into the callComp() wrapper, which allows me to call the comparator sub without segfaults, but I still cannot access the values either via @_ or $a/$b?

int callComp( SV* cmp, SV *a, SV *b) {
    int count, rv;
    GV *first = gv_fetchpv("a", TRUE, SVt_PV);
    GV *second = gv_fetchpv("b", TRUE, SVt_PV);

    dSP; ENTER; SAVETMPS; PUSHMARK(SP);

    XPUSHs( a );
    XPUSHs( b );

    PUTBACK; 

    GvSV(first)  = a;
    GvSV(second) = b;

    if( ( count = call_sv( cmp, G_SCALAR|G_NOARGS) ) != 1 )
        croak( "Bad comparator" );
        
    SPAGAIN; 
    rv = POPi;
    FREETMPS; LEAVE;
}
[download]

That said, when I look at what is done in the reduce() function in List::Util.xs, we are only scratching the surface yet. I've read perguts a dozen times over the last two days, but it's one of those documents that you have to experiment with each nuance of every sentance to understand the meaning and implications.

I was kind of hoping that Inline::C + perlapi would allow me to bypass the requirement for acquiring that understanding--but it seems not.

I'm not sure that I am interested in becoming an XS programmer. The only help available seems to be the p5p list--and they are all too busy doing, to be interested in teaching to do. Oh well!