Answer.

In Perl, hashes (associative arrays) are a powerful tool for storing key-value pairs. However, working with them requires caution, especially when simultaneously iterating and modifying the structure. Missing details can lead to errors that are difficult to diagnose.

History of the Question

Associative arrays were introduced in the early versions of Perl (Perl 1/2), making it one of the first languages with full hash support at the core level. Over time, additional features emerged: iteration with each, deletion (delete), mass transformation (map, grep), and dealing with changes in size/content during traversal.

Problem

Iterating over a hash and simultaneously modifying its contents, especially deleting elements, can lead to unexpected effects: skipping elements, re-visiting the same keys, or even an infinite loop. Moreover, the order of key traversal is not guaranteed and can vary between Perl versions.

Solution

Do not modify the hash during iteration when using each, as the internal cursor gets messed up.
For safe deletion of elements – first, gather the list of keys with keys, then iterate over it in a separate loop and delete.
Use while (my ($k, $v) = each %h) for normal iteration, but do not combine it with delete inside the loop if you want to avoid surprises.

Example of correct element deletion:

my %h = (a=>1, b=>2, c=>3);
for my $k (keys %h) {
    delete $h{$k} if $h{$k} == 2;
}

Example of incorrect approach:

while (my ($k, $v) = each %h) {
    delete $h{$k}; # This may lead to skipping keys
}

Key features:

The order of key traversal is not fixed and can change.
Iteration using each is sensitive to structural changes during execution.
For mass deletion, use traversal of a copy of the key list.

Tricky Questions.

Is it safe to delete elements from a hash inside a while (each %h) loop?

No, this may lead to skipping parts of the hash due to the reset of the internal iterator cursor.

What happens to the order of keys after deleting elements from the hash?

The order is not guaranteed and may change. Additionally, the order of traversal between programs on the same version of Perl may differ.

Can you change the value of a hash element during iteration through each?

Yes, changing the value (but not the structure) is safe.

Example:

while (my ($k, $v) = each %h) {
    $h{$k} = $v + 10;
}

Common Mistakes and Anti-Patterns

Deleting elements directly during each iteration.
Assuming the order of key traversal.
Modifying the structure of keys during traversal of the key array.

Real-life Example

Negative Case

Using element deletion through each in one loop:

my %h = (a=>1, b=>2, c=>3);
while (my ($k, $v) = each %h) {
    delete $h{$k} if $v == 1;
}

Pros:

Compact
Does not create additional arrays

Cons:

Risk of skipping elements
Unpredictable result

Positive Case

Creating a list of keys for deletion:

my %h = (a=>1, b=>2, c=>3);
for my $k (keys %h) {
    delete $h{$k} if $h{$k} == 1;
}