I've tried various "binmode" incantations -- you'd think that applying it to every filehandle in sight ought to be sufficient -- but nothing seems to be working. It even insists that /^\xFF\xFE/ is not a valid regexp, because \xFF\xFE isn't valid UTF-8. Well, no shit, perl; maybe that's because I'm not working in UTF-8, eh? It's a perfectly valid match for the beginning of a UTF-16-LE file.
And I can't find the sysadmin to ask him to roll back the change. And I really need to get some stuff done within the next day or so, and can't do it without either using my existing tools, or rewriting them to deal with whatever change has happened inside perl.
$#@*!
It should not be trying to tell me what encoding I'm working in. It should just treat the file as a byte stream, and leave encoding stuff up to me. Stupid clever language designer, thinking he's cleverer than me. :-P