As it happened, changing the script to use hashes made it considerably faster--so much so that it became almost as convenient to use as the C program. Actually, it became more convenient because I'd built better error-checking into it. The C program was content to try reading a binary file--of course, that'd generate an error--whereas the perl script refused to mess with binaries.

In that regard, at least, I judged the script "ok"; and because it was no longer grotesquely slow, I could even (cautiously) consider it "better." Of course, if the C programmer beefs up the error-checking, then his app wins again. :)