If Perl's function calls are slow, the right response is to try to fix the implementation, not to try to avoid function calls.

As someone with more than a passing familiarity with the internals of Perl 5, do you consider it feasible^* to substantially improve the current performance of function calls?

Update: ^*Given the nature of the forum, perhaps practical is a better word than feasible in this context.