Well, thats true I suppose but im not sure its entirely relevent, the tail recursion optimization only applies to subs which call themselves as the last statement. So insofar as the mutually recursive routines are tail recursive as well this optimization pattern still applies.

For those that dont know, (not you BrowserUK) tail optimization is really only _important_ in languages without looping constructs, in any language with looping constructs its just a nice bonus.

I simply meant that I don't see any easy way of avoiding the goto in these mutually recursive routines:

#! perl -slw
use strict;

my( $count1, $count2 );

sub a1 {
    b1( $_ [ 1 ] );
    goto &a2;
}

sub a2 {
    $count1++;
    $_[ 0 ]-- and goto &b2;
}


sub b1 {
    $_[ 0 ]-- and goto &b2;
}

sub b2 {
    $count2++;
    goto &a2;
}

my( $n1, $n2 ) = ( 100, 50 );
a1( $n1, $n2 );

print "c1: $count1; c2: $count2";

__END__
[15:37:55.16] P:\test>mutrec
c1: 151; c2: 150
[download]

This is not a particularly useful example, but imagine that one (pair) are reading a huge XML file for instance, and the other is validating it.

Rather than having to read the whole file, the first reads a bit and the second processes a bit and the first reads a bit more and so on. Both retain their own state across calls to the other. Think coroutines; a consumer and producer that each retain their running state, whilst switching control between themselves

---

Examine what is said, not who speaks.

Silence betokens consent.

Love the truth but pardon error.

If you're not afraid to call it like it is, then it sure is possible. It did take three refactors until I got it as clear as this, though:

#!/usr/bin/perl
use strict;
use warnings;

my( $count1, $count2 );

sub a1 {
    a2b2( --$_[ 1 ],  1 );
    a2b2(   $_[ 0 ], !1 );
}

sub a2b2 {
    $_[ 1 ] or goto a2;
    {
        b2: $count2++;
        a2: $count1++;
        $_[ 0 ]-- and redo;
    }
}

my( $n1, $n2 ) = ( 100, 50 );
a1( $n1, $n2 );

print "c1: $count1; c2: $count2\n";
[download]

Makeshifts last the longest.