Re: With the months of planning and the what-if scenarios
And that's the key point : months of planning.
Most customers that call me in want me to solve a given problem NOW. They don't have a test server, because that is a cost center.
So I end up doing the job as cautiously as I can, asking all the questions I can think of before committing code to a production server.
Even then, sometimes things go wrong. Generally it's when someone comes in and says "this thing is broken now", and I most of the time I can answer "you never told me about that".
The rest of the time it's an edge case nobody was aware of.
But I basically have to code a test script that modifies nothing, test the hell out of it and, when I'm satisfied it does the job, then I port the code to the actual production script.
I've had my share of stomach butterflies, but so far so good.