Un/pack is used to convert from human readable perl scalars to unreadable binary gibberish that makes all kinds of dings and bells and sounds when you print it to console. General guidelines are, if you have fixed length data, use un/pack. If you have variable length data, use a regexp. Un/pack has a little bit of variable length data parsing ability FYI so the guidelines aren't set in stone. For XML use a regexp (or better a proper XML parser), for C structs or things that are mostly squares if you open the file in a text editor, use un/pack. Un/pack is faster than a regexp for the same data if both can do the same thing. A regexp can not convert unprintable binary to printable numbers (but see tr
). For a perl program that takes no input from a network, a disk file, or a C function and only outputs human readable text, you don't usually need un/pack. If you see you need to write "$number = 0xF000 | 0x80;" you might need to use pack later on. In a perfect world, all input/output is done through Perl modules dedicated to that I/O medium, so you never need to call un/pack. In real life, thats not true, and then you have to call un/pack. Un/pack makes Perl the ultimate glue language.