By your reasoning I would predict that local should not work on a hash value - because local replaces a scalar with another scalar temporarily. But lo and behold, look at this!

use Data::Dumper;
%foo = qw(hello world);
print Dumper \%foo;
{
  local $foo{test} = "Just for now";
  print Dumper \%foo;
}
print Dumper \%foo;
[download]

Therefore if someone is willing to do the work and there is no ambiguity in the parsing, this feature could make it into Perl 5 and it wouldn't necessarily be a bad thing.

Incidental note. One cannot use local on a lexical. This is an utterly arbitrary limitation - the feature was once put in only to be taken out because it was felt that where it happened it was undoubtably a mistake.

Wrong how? By suggesting that foreach() could work if suitably innovated? I don't see how this could be wrong. You are looking at the current implementation of foreach() and local() and declaring it impossible, unfeasible, or wrong. If everybody looked at Perl this way, there would be no Perl 6, let alone Perl 2.

Please note that I didn't say the patch would be less than 100 lines long. Perl's evolution is entirely based on this sort of principle. People expecting something to 'just work'. It doesn't. People fix it so that it does. This is Perl. You should be familiar with this by now. :-)

You seem to be stuck on the definition of 'variable'. Where you and the original poster differ is that you don't see a[5] as a variable. I and the original poster *do* consider a[5] to be a variable. Fine, it doesn't have a simple name, but then, who cares about simple names? Your example showing that local() has a wider scope is wrong, since I am sure you should realize that:

package main;
local $b = 2;
$main::b = 3;
print "$b\n"; # prints "3\n"
[download]

Meaning, that '$b' is really just ${$main::{'b'}}, which is just as much a 'variable' or not a 'variable' as $a[5] is.

I think I've made my case... :-)

No, I am not looking at the current implementation of Perl. I am looking at a very fundamental principle of the syntax for computer language:

Any syntax should have a consistent meaning/interpretation, and can be consistently recognized by computer as a presentation of the same existance.