Rick Troth <[EMAIL PROTECTED]> shared an anecdote about code maint. I have my own:
Back in about 1982, at UofW, I wrote a tiny assembler routine to test a local DIAGNOSE. This unprivileged DIAG would return a userid's logon status and message flags, so you could tell if (s)he had MSG ON before writing a long msg (or sending an AYT?). Since it was just to test it, and was only a few lines, I didn't comment it, did no REGEQU, etc. Over the next few months, I kept tinkering, adding the ability to read a list of userids, output to the stack, etc., etc. Then someone put it on the Y-disk. Every summer when the Evil Students were gone, we'd run a local mod for a few weeks that monitored files read from the S- and Y-disks, so we could look at improving performance by putting stuff in the CMS nucleus (like COPYFILE, which we did long before IBM did). Summer of 1985, my MODULE was the most heavily used program on the Y-disk. At that point, it was over 2,000 lines of uncommented, un-REGEQU-ed assembler. I decided I'd better move to another country before something broke and I had to fix it... ...and since then, I've been more or less religious about commenting! ...phsiii