eval { require '/path/to/whatever.filter' };use Carp;
function whatever {
carp "oops";
# do filters and stuff
return $foo;
}
1;
and NOT die if there's an error in the included fileThat would kinda defeat the point of REQUIRE, though, wouldn't it.
That would kinda defeat the point of REQUIRE, though, wouldn't it.
$check = require '/path/to/whatever.filter';
if ($check) {
# do whatever
}# main script
use lib '/path/to';
use whatever;
$bar = whatever('example');
# /path/to/whatever.pm
package whatever;
use Carp;
sub whatever {
carp "oops";
// do filters and stuff
return $foo;
}
1;How many included files are potentially involved?
$result = do '/path/to/whatever.filter';
if ($result) {
require '/path/to/whatever.filter';
$check = defined(&whatever);
}
if ($check) {
# do whatever
}
# using Perl v. 5.22.0
use Time::HiRes qw(gettimeofday);
# Test 1: 0.196840047836304
$start_time = microtime(TRUE);
for ($i = 0; $i < 10000; $i++) {
do 'random.lib';
}
$end_time = microtime(TRUE);
print 'Test 1: ';
print ($end_time - $start_time);
print "\n";
# Test 2: 0.254052877426147
$start_time = microtime(TRUE);
for ($i = 0; $i < 10000; $i++) {
eval { require 'random.lib' };
}
$end_time = microtime(TRUE);
print 'Test 2: ';
print ($end_time - $start_time);
# main script
require '/path/to/whatever.filter';
$check = defined(&whatever);
if ($check) {
$bar = whatever('example');
}
# /path/to/whatever.filter
sub whatever {
$foo = false;
eval '
# do stuff
$foo = true;
';
return $foo;
}
1;
I THINK that putting EVAL inside of the included file will be faster, although it leaves an opportunity for an error.At this point you could also start thinking about the likelihood of various types of error. If the file-to-be-included doesn't exist at all, it would be a pretty lethal error--but not one that is awfully likely to occur, unless your .pl files really do have a habit of eloping with gremlins. Even if you did all this at 3 AM before crashing for the night, you can manage one quick test-run just to ensure you haven't, say, mistyped the name of the file. And then if the included file itself contains error handling, there should be a limit to how drastic the consequences of an error might be.
Even if you did all this at 3 AM before crashing for the night, you can manage one quick test-run just to ensure you haven't, say, mistyped the name of the file.
