We accept this as ordinary: error messages are nearly useless.
We don't even expect that the error given by some misbehaving system is accurate. I'd guess our confidence level in that hovers around 80% or so: it's pretty likely that whatever we got will eventually lead us to a solution, but there's no guarantee, is there?
We certainly don't expect that the error display will help us solve the problem. We've all been through Microsoft's near useless "troubleshooters" and we know they are NOT going to help us. No one else is any better. Maybe, if you are very, very lucky, the programs Help system might lead you to an answer, but we don't get outraged when it doesn't help us at all, do we?
Nope. We are conditioned to accept that error messages aren't particularly reliable or useful. When an error message actually is spot on, and tells you what you need to do to fix it, we are (quite naturally) mistrustful. In this case, the error gave the specific remedy, but the recipient went to the newsgroups, either to double check (and who can blame that instinct?) or because they barely even read the message at all, which is another habit that useless messages instill in us.
It's hard to write good error messages. After all, if the programmer EXPECTED this bad thing to happen, they'd usually do something to prevent it. Still, I think we could do better than we do. Programmers do have more than simple numeric messages to choose from, and if they are going to use numeric messages, should at least tell us whether this is a system error number (we could look that up for a clue; see Numeric Unix Error Messages) or something they made up. If they are using their own errors, give us a pointer to a web site that lists them: at least give us a chance.
Wouldn't it be nice to be suprised if an error message WASN'T helpful?
See Numeric Unix Error Messages also.
Got something to add? Send me email.
More Articles by Tony Lawrence © 2011-03-27 Tony Lawrence