In my previous life I was maintaining some backend services. One of the most common reasons for a service dying was a typo in a name in some rarely visited code path. I was tempted to write a similar module that would be a little bit smarter: when some identifier (variable, method, etc.) was not found, instead of raising an exception, try to find a similarly named thing and use that instead. I never did find a good way of implementing that.
I think you might like something like TheFuck?
EDIT: This bash commandline completion has been updated to by default wait for the user to verify the replacement command is what they wanted; this pretty much turns this from a terminal Russian Roulette into a kinda handy autosuggestion!
How about better testing and path coverage of your code? I hope that was also explored as a way to stabilize your codebase? Otherwise your suggestion does come offa bit as 'cowboy-ish' to say the least ...
That came from Smalltalk, where it was named doesNotUnderstand. It's always difficult to choose between crippling the language and encouraging stupidity.
