This is something of a pet peeve, so I'll reply. Let's say that the only stack trace entries TCO will eliminate are not only useless, but actually impede debugging. Let's say 'rf' is a recursive function. If it dumped core, without TCO you would see something like 'rf rf rf rf ...'. With TCO, you would see 'rf' once, just as if it dumped core while on a loop. Which is actually what TCO means: translation of recursive functions into functions that use loops, at the AST level. The result is code that's easier to debug (because it's not recursive at the stack trace level), easier to understand (because it's recursive at code level) and that performs better (because it's a loop at machine code level).

So yeah, you could make the case that it's not "an accurate stack trace", because it isn't. But, you don't want that. You want TCO. Trust me on this.

Preemptive strike for the bikeshedder crowd in the back :) : if you think you can deduce some useful fact from a recursive function that's stacktraced, like for example where in the loop the function gave an error, then you're wrong. Go learn to use a debugger.