Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!cs.utexas.edu!uunet!cbmvax!schmid
From: schmid@spica.cbmvax.cbm.commodore.com (Bob Schmid)
Newsgroups: comp.lang.perl
Subject: Efficiency of eval (emulating multidimentional arrays)
Message-ID: <13041@cbmvax.commodore.com>
Date: 5 Jul 90 01:35:42 GMT
Sender: news@cbmvax.commodore.com
Distribution: comp.lang.perl
Organization: /user/schmid/.organization
Lines: 36

I'm using eval to create a large number of associative arrays based on
different, variable dependent, names.  Eval seems to take about 1
millisecond extra per eval on a Sparcstation1 compared with explicitly
typing out the variable names.  This seems to me like a good tradeoff
between execution speed and manageable coding complexity (I really
need that extra array index!).

Any comments on eval efficiency or better ways to skin the cat will be
most appreciated....

The following sample program uses integer indexing.  I usually use
strings (hence the need for associative arrays).

for $cell (1..100)
{
    for $i (1..100)
    {
	eval("\$junk_$cell{\$i} = \$i");
    }
}
for $cell (1..100)
{
    while (($key, $val) = eval("each \%junk_$cell"))
    {
        print "\$key=$key \$val=$val";
    }
}

------------------------------------------------------
R. Schmid     <uunet!cbmvax!schmid>
              <schmid@cbmvax.cbm.commodore.com>

Commodore Technology
1200 Wilson Drive
West Chester, PA  19380
------------------------------------------------------