in reply to Efficient access to sparse lists?

There are a couple options. You need to know the index, so foreach is out. You could do something like this: 
for(my $x = 0; $x <= $#array; $x++) {
    next unless $array[$x];
    # do stuff with $x and $array[$x]
}

[download]

But that is obviously inefficient, especially if your list is VERY sparse. Like with elements numbered 100, 435, 1040, etc.

Why not use a hash? You can convert a sparse list to a hash like so:

my %hash = 
    map { $x => $y } 
    grep { my $y = $list[$x]; $y ne $undef; [$x, $y] }
    for(my $x = 0; $x <= $#list; $x++);

[download]

And use it like this:

$list{100} = 'dog';
$list{435} = 'cat';
$list{1040} = 'tax form';

while(($index, $element) = each %list) {
    # do stuff with $index and $element
}

[download]

Other alternatives include using a pseudohash, or writing a tied array implementation that acts like a list in your code but works like a hash behind the scenes.

Good luck!

Replies are listed 'Best First'.
(corrections) Re: Answer: Efficient access to sparse lists?
by mwp (Hermit) on Jan 25, 2001 at 16:57 UTC
    Instead of the map/grep/for mess I gave you, try something like this to convert a sparse list to a hash: 
    my %hash = ();
for(my $x = 0; $x <= $#sparse_list; $x++) {
    my $y = $spase_list[$x];
    next if $y eq undef;
    $hash{$x} = $y;
}

    [download]

    As an aside, you can use exists and delete on array elements but it comes highly discouraged. So don't do it.

    Apologies for the typos...
[reply]
[d/l]
      The problem is that I need fast *ordered* access, something that a hash really won't help me with.
[reply]

        Ah, now I finally understand what you want. (:

        You need a more complex data structure. You can wrap a nice interface around it if you like, but I'll leave that part up to you (or others).

        I think you've found a reason to implement a linked list in Perl (normally you don't need such things in Perl). I went with a circular, doubly-linked list for ease of implementation (where "turn 0" is a dummy turn used to find the first and last turns): 

        my %turn= (
    0=>{Next=>5,Prev=>53},
    5=>{Turn=>"turn 0",Prev=>0,Next=>28},
    28=>{Turn=>"turn 28",Prev=>5,Next=>53},
    53=>{Turn=>"turn 53",Prev=>28,Next=>0},
);
sub pushTurn {
    my( $num, $turn )= @_;
    my $last= $turn{0}{Prev};
    $turn{$num}= {Turn=>$turn,Prev=>$last,Next=>0};
    $turn{0}{Prev}= $num;
    $turn{$last}{Next}= $num;
}
sub deleTurn {
    my( $num )= @_;
    my( $prev, $next )= @{ $turn{$num} }{qw( Prev Next )};
    $turn{$prev}{Next}= $next;
    $turn{$next}{Prev}= $prev;
    delete $turn{$num};
}

        [download]

        I hope that is close enough to correct that you can get an implementation working. {-:

        Use: 

        my %turn= ( 0=>{Next=>0,Prev=>0} );

        [download]
        to bootstrap the data structure.

                - tye (but my friends call me "Tye")
[reply]
[d/l]
[select]
        Have you considered a linked list in a hash?

        Update: Apologies for the redundant post - I hadn't discovered the notes depth setting until a few moments ago, so hadn't noticed tye's post.

        Pete

[reply]
        What you are talking about sounds like what a BTree is for. You can get it done in Perl, use DB_File (or the next generation and possibly immature BerkeleyDB) to get it in C.

        If you go with the latter two options watch out for the need to set a sort order. You will run into that will the all-Perl solution as well, but there you can modify the module to work that way...

[reply]

In Section Seekers of Perl Wisdom