I thought that in your code, if $empty_hash{foo}{bar} did not exist, then print "'foo'->'bar' exists\n"; would not be executed.
Did you, in fact, try to run the code? You'll see that that print statement indeed will not be executed.
As for the workaround you suggest, no. Checking whether a certain key exists in a hash is not a workaround for checking whether a subkey exists somewhere deeper down you structure. Obviously if the key doesn't exist, this implies that the subkey doesn't exist, but if the key existst this tells us nothing about the existance of the subkey. Proof:
use strict;
use warnings;
use Data::Dumper;
# Some arbitrary data
my %earth = ();
$earth{wind}{fire} = "water";
# Show me what we've got
print Dumper \%earth;
print "\n\n";
print "\nEarth has wind, wind has fire.\n" if exists $earth{wind}->{f
+ire};
delete $earth{wind}{fire};
print "The fire is extinguised.\n" unless exists $earth{wind}
+->{fire};
print "Yet earth still has wind.\n" if exists $earth{wind};
delete $earth{wind};
print "But the wind blows away.\n" unless exists $earth{wind}
+;
print Dumper \%earth;
print "\n\n";
print "Relight that fire!\n" if exists $earth{wind}{fir
+e}; # Autovivication happens here!
print Dumper \%earth;
print "\n\n";
$VAR1 = {
'wind' => {
'fire' => 'water'
}
};
Earth has wind, wind has fire.
The fire is extinguised.
Yet earth still has wind.
But the wind blows away.
$VAR1 = {};
$VAR1 = {
'wind' => {}
};
The earliest moment where you autovivify elements in %flag_assignments is the print statement in this snippet:
if (defined $flag_assignments{$temp_key}) {
print "KEY [$key] TK [$temp_key] TS [$temp_semester] FAS [$fla
+g_assignments{$temp_key}{starrez}] FAA [$flag_assignments{$temp_key}{
+abbreviation}]\n";
}
And you have many more statements like that where you autovivicate in string interpolation. This is not in itself a bad thing, but it explains the behaviour you are seeing.
|