[plt-scheme] two tool ideas

> It depends on the evaluation strategy.  The runtime could first
> allocate
> an uninitialized cons cell, and evaluation of the arguments could then
> "return" into the fields of the cell.  In this case, the second
> argument
> of the cons will sometimes behave as if in tail position.  Wadler's
> "listless" work deals with optimizations for such programs, which are
> said
> to be "tail-recursive modulo cons".

This topic recently came up on the Ocaml list.  The proposal was to
modify Ocaml's append and similar operators to use a tail-recursive mod
cons strategy so they'd consume less space.  I'd guess MzScheme's append
and friends could benefit as well.

-- Paul