I found this in Mail::IMAPClient. Where does the $_
in $SEARCH_KEYS{ uc($_) }
come from?
sub _quote_search {
my ( $self, @args ) = @_;
my @ret;
foreach my $v (@args) {
if ( ref($v) eq "SCALAR" ) {
push( @ret, $$v );
}
elsif ( exists $SEARCH_KEYS{ uc($_) } ) {
push( @ret, $v );
}
elsif ( @args == 1 ) {
push( @ret, $v ); # <3.17 compat: caller responsible for quoting
}
else {
push开发者_运维技巧( @ret, $self->Quote($v) );
}
}
return @ret;
}
That looks to me like a typo where the author converted an anonymous for loop foreach (@args)
to one with an explicit iterator variable foreach my $v (@args)
and forgot to convert all the incidences of $_
to $v
.
You should probably file a bug against the distribution on CPAN.
Even though this is probably a bug, lets consider how this code behaves.
The value of $_
will be determined by the current dynamic scope. What this means is that $_
will have whatever value (the dynamically scoped copy of) $_
has in the calling subroutine.
So for example if I have:
for (1 .. 5 ) {
foo();
bar();
}
sub foo {
print "\$_ = $_\n";
}
sub bar {
for ( 'a' .. 'c' ) {
foo();
}
}
You get output like:
$_ = 1
$_ = a
$_ = b
$_ = c
$_ = 2
$_ = a
$_ = b
$_ = c
...
It gets a little weirder in Perl 5.10 and up, where a lexical $_
exists.
for (1 .. 5 ) {
foo();
bar();
}
sub foo {
print "\$_ = $_\n";
}
sub bar {
my $_;
for ( 'a' .. 'c' ) {
foo();
}
}
Run this and get:
$_ = 1
$_ = 1
$_ = 1
$_ = 1
$_ = 2
$_ = 2
$_ = 2
$_ = 2
As you can see, if this isn't a bug, it's probably a bad idea.
精彩评论