10.1 Return-point syntax

[acarrico]

Shivers/Fisher 2006 pg 30:

"A third possibility borrows from SML's "value restriction:" restrict return points to be
either expressions or variable references (Milner et al., 1997). Variable references are
useful ret-pts for real programming, as they give the ability to name and then use "join
points" in multiple locations. This is clearer and more succinct than the awkward alternate
of binding the join point to a name, and then referring to it with -expanded return points
in the desired locations."
"Restricting return-point expressions to expressions and variable references eliminates
code blowup in transformations, since large ret-pt expressions can be let-bound and re-
placed by a name before replication. It eliminates issues of control effect, since both forms
of expression can be guaranteed to evaluate in a small, finite amount of time. For a real
programming language, this is the syntax we prefer."