Thu Feb 13 06:33:39 EST 2020

redo: providing temp space to imperative code

The thing that really strikes in redo is the way in which an
imperative update function is given the "ownership" of the output

Basically, it can do what it wants with that piece of storage while it
is running.  It is very nice to have this high level guard around a
dirty, in-place imperative update.

Write a C version?
Since I have the example, what would that look like?