Apart from the comic effect of the now well politically incorrect Little Britain, whilst recently trying to correct an issue I had with some company website, ( now fixed thanks ), I remember times in my Geekhood when it seemed impossible that the computer was right and I was wrong. Every programmer knows the feeling like they did everything right, and it was working a moment ago and it come up with what seems like a bug. You look at the code for the obvious errors, not there. You roll back a version, either that works for not, but as you put back the changes it still won’t work. After to some struggle you have to admin to defeat and ask the guy or gal in the next desk to take a look for you. Often as you explain the problem you talk yourself into the answer, or maybe your desk mate can say “oh they changed the compiler last week” , or “you’re running the old version, idiot!”. A fresh pair of eyes. Often where you are looking isn’t where you should look at all. But sometimes, just sometimes, you can’t crack the problem. Random errors are the worse. Works then doesn’t, then works again. There is a sort of rule of thumb.
Blame the user – usually yourself .
Blame the data – check the inputs and outputs, eyeball the data .
Blame your code.
Blame the library.
Blame the compiler – now I’ve seen this happen a couple of times. I was the guy in the next desk and I was shown some Cobol that should have worked but it threw an error at runtime. Now I don’t know how I knew this, maybe I read to many manuals, but I saw the subroutine was being call from an address which was not on a word boundary. An easy mistake to make, but hard to spot.
Blame the operating system. – maybe the disk is full, or there’s a known bug in windows, or a quirk in Chrome, these things happen, or a display driver, but not often.
Blame the hardware – You can check this by running on another machine.
Then you’re stumped.
I’ve only been stumped twice in my life. At a total loss. But hello, hang on, I don’t like to be beaten!
I don’t know if I came up with the answers probably not, but here are the situations.
A mini computer for a hotel accounts dept kept going wrong. Crashes, corrupt memory, random errors.
They changed the disks, Upgraded the OS. Still no fix. They manufacturer changed all the memory. Still no go. They changed the machine, still no go. They gave up and flew back to Texas.
I was a junior programmer and was sent to spend Christmas in a nice hotel in London so I could be a face to blame when it went wrong. It didn’t go wrong. Once. I went home, it went wrong again.
Answer. The office was on the second floor. There was a butcher on the first floor with big meat cutting saws. Over Christmas he didn’t work. The computer did. The solution was to build a Faraday cage around the machine room. Solved.
Second case. An international airport had a similar problem. Different computer company. Worked and then didn’t, but it always seemed to fail on Thursdays. Again environment. They had a training radar school on Thursdays. Easy fix. aluminum foil in the case ( so they told me ).
So if you’re ever stumped take comfort in the fact that cosmic rays can knock electrons out of chips. This is rare, but with so many computers in the world, this just might be the excuse you’re looking for.