Map reverse surprise

[philiprbrenan]

Please provide a compile time message explaining that map{reverse $_} does not behave as one might immediately expect as shown below:

#!/usr/bin/perl
use warnings FATAL => qw(all);
use strict;
use Data::Dump qw(dump);
use feature qw(say);

say dump map{            $_} 10..19; # (10 .. 19) # Expected
say dump map{    reverse $_} 10..19; # (10 .. 19) # Surprising
say dump map{scalar reverse} 10..19; # ("01", 11, 21, 31, 41, 51, 61, 71, 81, 91) # Annoying

[richardleach]

What behaviour might people otherwise be expecting? (The results of the examples do seem to follow the documentation for mapand reverse.)

[Grinnz]

This is a general problem with two sources:

• reverse not reversing strings in list context
• list context being the default in unexpected places, even more common than this is hash construction and subroutine arguments

More warnings are helpful but I'm not sure exactly what we could warn about in this specific case.

[gugod]

The way I see it... the surprising factors are (1) when reverse itself is in list context, and (2) when reverse is passed only one argument. Such as these cases:

# a single string literal
($foo) = reverse("foo");

# a single scalar variable
%ridx = map { reverse($_) => 1 } ("foo", "bar");

When both (1) and (2) are satisfied, the list-reverse is essentially an no-op -- it does effectly nothing to the list, nor to the only element inside.

A warning message can probably be yield, saying: "reversing a list of one item would have no effect. Do you mean: scalar reverse($x) ?"

But there should probably be no warnings when an array of just one item is given to the list-reverse

@x = ("foo");
my @y = reverse(@x);