Let's say there are three paths in @INC
: path1
, path2
and path3
. Under each of these paths, there is a module named foo.pm
. If I now load foo.pm
in my script via use foo;
, which of the foo.pm
s is actually going to be loaded? Or in other words, what is perl's search order for paths in @INC
?

- 9,586
- 6
- 49
- 79

- 40,802
- 41
- 107
- 134
-
1possible duplicate of [How do I choose a package name for a custom Perl module that does not collide with builtin or CPAN packages names?](http://stackoverflow.com/questions/658955/how-do-i-choose-a-package-name-for-a-custom-perl-module-that-does-not-collide-wi) (That question covers more ground, but it includes this question.) – cjm Dec 02 '11 at 16:57
-
possible duplicate of [How does a Perl program know where to find the file containing Perl module it uses?](http://stackoverflow.com/questions/2526520/how-does-a-perl-program-know-where-to-find-the-file-containing-perl-module-it-us) . Also see http://stackoverflow.com/questions/2526804/how-is-perls-inc-constructed-aka-what-are-all-the-ways-of-affecting-where-pe – DVK Dec 03 '11 at 09:02
2 Answers
perldoc -v %INC shows which path was chosen:
use Data::Dumper;
print Dumper(\%INC);
Or...
perl -Mfoo -e 'print $INC{"foo.pm"}'
require shows some psuedo-code which implies the search order:
foreach $prefix (@INC) {
}
Thus, path1
would be searched first.
path1, path2, path3. And perl will load path1/foo.pm
.
Why would you expect it to be any other?
Looking at perlfunc perlvar
, I can see that they don't explicitly say this, but they do say:
The array @INC contains the list of places that the do EXPR , require, or use constructs look for their library files.
I think the hint on there is list. It's unexceptional to expect a list to be processed first-to-last.
You could probably put this code, right before your use foo;
statement:
BEGIN { say "\@INC=(${\join( ', ', @INC )})"; }
If that still shows you @INC=(/path1, /path2, /path3)
then put this after the use
statement:
BEGIN { say "\$INC{'foo.pm'}=$INC{'foo.pm'}"; }
And if that one still shows $INC{'foo.pm'}=/path3/foo.pm
, then I think you're not specifying your search paths as well as you might. You might think you have foo.pm
in the same directory specified as '/path1', but the likelihood is that you've got some path messed up.

- 29,660
- 2
- 47
- 102
-
\@Axeman: I print @INC and the printed order is path1, path2 and path3, me too expected that it's path1/foo.pm is loaded, but in my case path3/foo.pm is loaded. there are two possibilities here, one is that the printed order is the reverse of search order, or something crazy... – Haiyuan Zhang Dec 02 '11 at 14:18
-
1@HaiyuanZhang `@INC` should print out in order, again as it is a list. Perhaps you do some manipulation with `@INC` in something that runs after the `BEGIN` blocks (and remember that `use` is an implicit `BEGIN` block). So you can write code like this: `unshift @INC, qw;use foo;` and end up with an implementation of `foo` in the path3. Sequential location in the file does not dictate sequential processing, necessarily. See [`use`](http://perldoc.perl.org/functions/use.html) – Axeman Dec 02 '11 at 14:24