First of all, ++ to you, simply for being (probably) the only other monk here that uses perl for X12. I'm just curious, which X12 transactions do you work with? I, coming from healthcare, do a lot with 270/271, 835 and 837. Personally, I use a somewhat more complicated library (written mostly by me) for handling X12's. So much fun.

Anyway, as to your particular implementation: I've found that some EDI sources use (almost certainly non-compliantly... but the spec has some minor vagueries) an "\r\n" pair as their segment separator, or, even wackier include a following linefeed (and/or carriage return) after their segment separator (since this is how the spec examples tend to make the files appear). So I end up with using:

($isa =~ /.{105}(.?\r?\n?)/)[0]
[download]

as my segment separator.

P.S. In other geeky stories... have you ever had to have an argument with your boss about why "grep" is inappropriate for X12 files?

• "Grep would be faster than using perl regexps."
• "But it's not line-based! And why should I subshell to grep, when I'm already in a perl process?"
• "Well... you can change it to use grep later."
• "ARGH!"