On Mon, Dec 09, 2002 at 10:27:20AM +1100, Damian Conway wrote:
> Nicholas Clark complained:
Yes. :-)
> No more than any of these are:
>
> key => "value";
> $hash{ key }
> qw( unbare );
>
> In each case, a sequence of \w characters is marked as a string in some way.
> Sure, it's not with the traditional '...' or qq(...), but it's a "Here Be
> Stryngs"
> marker just the same.
I don't think I'd include qw in that list, unless qq, qx, qr, m and the other
"quoting" constructions go in.
On Mon, Dec 09, 2002 at 12:34:52AM +0000, Kate L Pugh wrote:
> Nick wrote:
> >>> But barewords are not allowed under strict. So why is -bareword
> >>> being allowed?
>
> Damian wrote:
> >> Because it's not a bareword. ;-)
>
> Nick wrote:
> > I think it's bad documentation - that last sentence should not use the
> > identifier 'bareword' as its example identifier.
>
> I think that's the point of using that identifier - that something
> which would otherwise be a bareword is magically de-bareworded by the '-'.
OK. So I think I understand. Tokens that match /[A-Za-z_][A-Za-z_0-9]/
are considered barewords if they fail to be anything else. So there are at
least 7 that I'm aware of:
#!/usr/local/bin/perl
#use warnings;
use strict;
package Foo;
package main;
sub bar {
}
use vars qw (%hash @array);
# 1 Subroutines that have been seen
bar;
# 2 Packages that have been seen
my $foo = Foo->can ("thing");
# 3 Left of a =>
%hash = (key => 'value');
# 4 Inside {}
$foo = $hash{key};
# 5 perl1-ism for hashes
($foo) = keys hash;
# 6 perl1-ism for arrays
$foo = pop array;
# 7 string negation
$foo = -identifier;
print "$foo\n";
# Not sure - are file handles barewords?
print STDOUT "";
# Are format names?
format STDOUT =
Done
.
write;
__END__
Which runs with this output.
-identifier
Done
Nicholas Clark
