On Thu, 28 Feb 2002 19:15:16 -0500, Joshua Juran wrote:
>If you wrote this as a reuseable subroutine, $origpath and $destpath would
>vary between invocations.
Yes. But only $origpath matters.
>To avoid the performance hit of recompiling the
>regex each time, you could use eval like this:
>
>my $orig2dest = eval {
> sub {
> my ($file) = @_;
> (my $dest = $file) =~ s/^\Q$origpath/$destpath/;
> return $dest;
> }
>};
Wrong eval() here! This *does *not* recompile the sub when you're
creating a new sub! For that, you need eval STRING!
A few remarks:
* a closure would be a neat idea here. For that, you create a sub with
$origpath and $destpath as parameters, and which returns a sub -- this
sub.
* For older Perls, such as the old MacPerl, the /o option is useful.
* Unfortunately, with these older perls, /o and closures don't work
well together.
* So, for these older perls, you need an eval string in the closure
producing sub.
* For newer perls, such as the brandnew MacPerl, Niether the /o option
nor the eval are really necessary. Regexes based on a variable will only
recompile the regex if the variable has changed between invocations.
So, for the sake of old MacPerls:
sub genconversion {
my($origpath, $destpath) = @_;
return eval <<'-END-';
sub {
my ($file) = @_;
$file =~ s/^\Q$origpath/$destpath/o;
return $file;
}
-END-
}
$orig2dest = genconversion('/foo', '/bar');
It should work on newer perls too, although it's a bit of overkill to do
it this way.
>Then, replace the substitution line in the original code with:
>
>my $dest = $orig2dest->($file);
You got that right.
--
Bart.
