Sun Aug 16 18:46:55 CEST 2009

Linear cons cells

It might be best to switch to an explicit representation for linear
CONS cells.  Especially for the interaction with Scheme, it might be
confusing to have the same datastructure represent two different

The hack is of course that inside the PF primitives, it's possible to
use direct calls to read-only ex_ functions, without the need to make
them polymorphic to both linear and nonlinear lists.

Anyways, this seems something to postpone until it becomes a problem.
It should be straightforward to do: it's mostly the functions
_pf_copy_from_graph and _pf_copy_to_graph.