http://qs321.pair.com?node_id=1142542


in reply to Smoothsort

Hello QuillMeantTen, and welcome to the Monastery!

Since this is a learning exercise, I want to comment on your use of C-style for loops. For example, from sub heapify:

for(my $i = $start; $i < leonardo($order);$i++){ print "$$tab_ref[$i]\n"; }

As a general rule, whenever you have a C-style for loop in which the reinitialisation expression increments a variable by one, you should re-cast it as a Perl-style foreach loop instead:

for my $i ($start .. leonardo($order) - 1) { print "$$tab_ref[$i]\n"; }

Or, even better, put it on a single line by making the foreach a statement modifier:

print "$$tab_ref[$_]\n" for $start .. leonardo($order) - 1;

— which has the added advantage of simplifying the code by removing a variable declaration.

Not only is a Perl foreach loop intrinsically faster than its C-style equivalent; in this case you get a significant extra benefit in efficiency, because the subroutine call to leonardo($order) is made only once. Here’s a longer example to emphasise the point:

use strict; use warnings; my $x = 5; for (my $i = 0; $i < bar($x); $i++) { print "for loop: $i\n"; } print "----------\n"; for my $j (0 .. bar($x) - 1) { print "foreach loop: $j\n"; } sub bar { my $y = shift; print "bar($y)\n"; return $y; }

Output:

19:23 >perl 1378_Smoothsort.pl bar(5) for loop: 0 bar(5) for loop: 1 bar(5) for loop: 2 bar(5) for loop: 3 bar(5) for loop: 4 bar(5) ---------- bar(5) foreach loop: 0 foreach loop: 1 foreach loop: 2 foreach loop: 3 foreach loop: 4 19:23 >

Of course, this works only because the value returned by leonardo($order) doesn’t change (because $order isn’t changed in the body of the loop).

Another point to note: recursive functions like sub leonardo can often be significantly streamlined by memoization. This is explained in Chapter 3 of Higher-Order Perl by Mark Jason Dominus, which is available for free from http://hop.perl.plover.com/. See also the module Memoize (also by MJD).

Update: Fixed off-by-one error in foreach loops, thanks to AnomalousMonk (twice!).

Hope that helps,

Athanasius <°(((><contra mundum Iustus alius egestas vitae, eros Piratica,

Replies are listed 'Best First'.
Re^2: Smoothsort
by Laurent_R (Canon) on Sep 20, 2015 at 08:44 UTC
    Hi Athanasius,

    your point about the range-computing subroutine being called only once when using a Perl-style foreach loop (as compared to the "equivalent" C-style loop) is very interesting.

    I sometimes wondered, in similar cases, whether I should pre-compute the range limits in order to sort of cache them before entering the for loop, but never actually cared to check, now I know that, in such cases, Perl is actually doing the value caching for you. Good to know. Thank you, Athanasius.