Yes you are completely correct in your scenario. Its much better to use regex when unpack can be often thrown out because of changing length...

One question : Do you embed your regexes into code, or you put them as variables?

I have never written PERL code that has run 1/3rd at the speed of similar C code - usually the values I get are around 1/20th median. Best I got was 1/10th.

But I believe you. I had also read an article (but now I forget its contents and URI) that showed how we can optimize PERL by giving it hints and making appropriate use of functions.. I wold post a thread soon on this asking for more information... do you for example have any? (For eg, favouring unpack over substr etc)

I frankly believe GUI makes a very bad case for OO (Object Oriented is what I mean), but yes, many things can be done better without OO, specially in cases when a UML digram was its predecessor.