Re: Attempting to use $_ in "sub" inside "map" (updated)

[AnomalousMonk]

One way, albeit not pretty, is to form a closure | lexical closure (if that's the proper terminology) over the value of the global $_ and use the closed-over value:

Win8 Strawberry 5.8.9.5 (32)  Sat 05/08/2021  0:53:55
C:\@Work\Perl\monks
>perl
use strict;
use warnings;
my %dispatch = (
    normal => sub{"$_[0] Normal dispatch"},
    map {
        my $__ = $_;
        $_ => sub{"$_[0] Sub returns $__"}
        }
        qw| Uno Dos tres|
    );

print $dispatch{"normal"}->(0),"\n";  # 0 Normal dispatch (as expected
+)
print $dispatch{"Uno"}   ->(1),"\n";  #  WANT: "1 Sub returns Uno"
print $dispatch{"Dos"}   ->(2),"\n";  #  WANT: "2 Sub returns Dos"
^Z
0 Normal dispatch
1 Sub returns Uno
2 Sub returns Dos

Update 1: Also see this article on closure.

Update 2: Another way, not involving closures - although closures are not to be scorned!

Win8 Strawberry 5.8.9.5 (32)  Mon 05/10/2021  9:33:38
C:\@Work\Perl\monks
>perl
use strict;
use warnings;

my %dispatch = (
    normal   => sub{ "$_[0] Normal dispatch" },
    map { $_ => eval qq{ sub{ "$_ returns \$_[0]" } } } qw(Uno Dos tre
+s),
    );

print $dispatch{'normal'}->(  0), "\n";
print $dispatch{'Uno'   }->(  1), "\n";
print $dispatch{'Dos'   }->( 22), "\n";
print $dispatch{'tres'  }->(333), "\n";
^Z
0 Normal dispatch
Uno returns 1
Dos returns 22
tres returns 333

[NetWallah]

Thank you (++)! - that works well.

[LanX]

> is to form a closure | lexical closure (if that's the proper terminology)

I'm pretty sure both are correct. I have trouble to imagine closed over variables which are not lexical. (I usually call those global ;-)

Personally I'd use "lexical closure" to disambiguate, if there is another interpretation of "closure" in the context.

The same way we use "file-glob" and "type-glob" to distinguish between glob and the * sigil

[AnomalousMonk]

I have trouble to imagine closed over variables which are not lexical.

Yeah, I was thinking about it after posting and I couldn't come up with an example of non-lexical closure either. I'll just add a link to the Wikipedia article and let everyone else figure it out. :)

[tobyink]

In languages without lexically scoped variables or something approximating them, there's not really any such thing as a closure. You can still have first-class functions (i.e. coderefs); they just have no need to close over anything.

[LanX]

> first-class functions (i.e. coderefs)

Doesn't equal closures. OK PHP is using that terminology, but it's wrong ...

... anyway who's using PHP anymore ;-)

> In languages without lexically scoped variables or something approximating them

do you have an example for such a language with functions as first class objects?