The localization behavior, in my view, is expected and a benefit, not a wart. The only surprise was the localization was transparent and legal, but again, to me, I never would have found out on my own because repurposing variables like that is code smell.

Maybe it is a code smell...

But what I was trying to do is to find the smallest number in a certain range satisfying some condition and then go on using that found value.

Doing it like

my $i;

for $i (1..10) {
  last if &condition($i);
}

&do_more_stuff_with_the_value_found($i);
[download]

Looked (at the time) to be a straightforward solution and I was suprised that it did not work.

This is just for explaining where this comes from (I am not a really good programmer I am afraid...)

And even accepting that this is not the way to do it (there are of course many other ways), it is neverthless surprising that this actually does work:

my $i;

for ($i=1; $i<=10; $i) {
  last if &condition($i);
}

&do_more_stuff_with_the_value_found($i);
[download]

This is to my eyes at least an inconsistency (and thus a wart). But whatever, this is how it was, is and will be...

That use case is completely sensible. Your original was a string overwritten with a count and reminded me of the awful code I've had to deal with over the years where there is one $i or $output and friends shared around huge scopes or even more than one script, concatted or overwritten willy nilly. Your use case in a small scope seems totally kosher though it's outside my style or sensibility.

I can remember that long time (say: 30 or more years) ago, I also wrote loops like that.

But now that you mention it, and thinking about it, I notice always write those kind of loops like:

my $interesting_index= 'not existing';

for ($i=1; $i<=10; ++$i) {
  next unless &condition($i);
  $interesting_index= $i;
  last;
}
# spit an error if $interesting_index is still 'not existing'
[download]

The reason behind this is simply that your loop might not have found a valid value.

---

s$$([},&%#}/&/]+}%&{})*;#$&&s&&$^X.($'^"%]=\&(|?*{%
+.+=%;.#_}\&"^"-+%*).}%:##%}={~=~:.")&e&&s""`$''`"e

I have made this mistake in the past. I now usually avoid the issue by using a function such as "first_index" in List::MoreUtils.
Bill