[johanbev]

Most heavily optimizing compilers do strange things to the code. You wouldn't want to debug an -O3 program, just as you probably wouldn't want to debug with TCO turned on.

Luckily, in the lisp and scheme families (possibly also other image based PLs), it's easy to mix compiled (and optimized) and interpreted code. Whenever i run into a bug such that i need the debugger to fix it, I always replace the compiled function with an interpreted version first, making it much easier to debug. What you get in the stacktraces is the same as you read on the screen.

In most lisp implementations, this maneuver can be performed completely on line, you don't have to restart the program or recompile anything but the function under inspection itself. When the runtime signals an exception or fault, just tell lisp to interpret the suspect function. Then move up the stack to the function that called the suspect, and restart that stack frame et voilá, bob is your uncle.