[plt-scheme] build-list for large lists

On Apr  9, Matthias Felleisen wrote:
> 
> Eli and I have had this long-running discussion. As Kent and Bob
> Hieb showed to me two decades ago, plain recursion usually wins in
> the context of an optimizing compiler because the compiler knows
> more about managing stacks and stuff than programmers do.
> 
> I for one am happy to see that Eli is now wrong in the context of
> mzscheme :-)

Just to clarify "now" started probably more than a year ago, when I
stopped writing iterative loops to get more speed.

Iteration+`reverse!' was much faster for a really long time, then just
faster (not "much") since 3m, and then `reverse!' went out leaving
iteration+`reverse' or recursion at roughly the same speed.  Together
with improvements to the GC and the JIT, the recursive versions became
increasingly faster, while the iterative ones got left behind.

-- 
          ((lambda (x) (x x)) (lambda (x) (x x)))          Eli Barzilay:
                  http://www.barzilay.org/                 Maze is Life!