“Inconvenient,” perhaps, but not “useless.”

In all seriousness, what is so utterly wrong with just using a subdirectory?   Apple’s OS/X system is rather elegant, I think, in taking that approach with how they package all applications:   they’re actually folders.   Just put your executable together more-or-less as a stub, and put the necessary Perl source modules into a directory in a chosen location.

(When I have done this, I used a nested directory structure:   the outer directory contained README.TXT and an inner folder named guts.   The entire thing was marked read-only.   And the README file contained a message which basically said:   “Keep your mitts off that folder and everything that it contains.   These aren’t the ’droids you’re looking for.”   I got a few whimsical comments about that, so I know that some people read it.)   To me, this is every bit as convenient a form of “packaging” as a DLL-file would be, and considerably more maintainable.   It might well be more efficient, too, because Windows will routinely use VM-mapped file I/O when working with files.