(jeffa) Re: Efficient Rating of Test Answers

Ughh! No matter how you slice it, that's an icky problem. I try to avoid those nasty if-elsif-else statements with a hash table. japhy calls it a 'Dispatch Table', which sounds like a great name to me. The idea is that the keys in the hash point to subroutines:


my %dispatch = (
     g6 => \&add_one,
     g7 => \&add_one,
     g8 => \&add_one,
     g9 => \&add_one,
     g0 => \&add_half,
     g1 => \&add_half,
     g2 => \&add_half,
     g3 => \&add_half,
     g4 => \&add_half,
     g5 => \&add_half,
     k0 => \&add_half,
     k1 => \&add_half,
     k2 => \&add_half,
     k3 => \&add_half,
     k4 => \&add_half,
     k5 => \&add_half,
);

sub add_one {
    my $total = shift;
    $total += 1;
    return $total;
}

sub add_half {
    my $total = shift;
    $total += 1;
    return $total;
}
[download]

And you access the table like so:


my $total = 10;
$total    = $dispatch{'g6'}->($total);
$total    = $dispatch{'k5'}->($total);

print $total, "\n";
[download]

This is not a complete solution for you, but it is a start.

----------------------------------------------------
 perl -le '$x="jeff";$x++ for(0..4482550);print $x'
----------------------------------------------------

Comment on (jeffa) Re: Efficient Rating of Test Answers Select or Download Code

Replies are listed 'Best First'.
Re: (jeffa) Re: Efficient Rating of Test Answers by Hofmator (Curate) on Aug 14, 2001 at 13:37 UTC
You are describing the general concept of a 'Dispatch Table'. While very useful at times I think that's a little overkill here. Your subroutines do exactly the same thing - add a number - and only this number differs. So why not just store the number as proposed by VSarkiss. Your code would then simplify to `my %lookup = ( # no longer a dispatch g6 => 1, g7 => 1, g8 => 1, g9 => 1, g0 => 0.5, # ... ); my $total = 10; $total += exists($lookup{'g6'})? $lookup{'g6'} : 0; print $total, "\n";` [download] I think I still prefer the regex solution by enoch in this easy case where the keys/answers are so 'handy' to match. -- Hofmator	[reply] [d/l]
(jeffa) 3Re: Efficient Rating of Test Answers by jeffa (Bishop) on Aug 14, 2001 at 19:16 UTC
Very nice! How about generating the table with Perl instead of a text editor? I'll just borrow the regexes you and enoch provided, if you don't mind :D `my %lookup; foreach my $alpha ('a'..'k') { foreach my $numb (0..9) { $_ = $alpha . $numb; $lookup{$_} = /^g[6-9]$/ ? 1 : /^[gk][0-5]$/ ? .5 : 0; } }` [download] however, this is not very efficient if your memory is tied behind your back. ;) ---------------------------------------------------- perl -le '$x="jeff";$x++ for(0..4482550);print $x' ----------------------------------------------------	[reply] [d/l]