Sun Dec 6 10:37:42 CET 2009
To simplify compilation from Scheme -> C using only the LEAF object
model and manual memory management, all objects need to become leaf
objects (i'm looking at integers and characters), so at least there is
a unified model for data and operations at a lower level.
As became quite clear toying with the Java JNI, having to special case
data items that are not objects is quite a pain.
It's important not to forget the goal of the libprim project: have
low-level code and data models that can be _augmented_ with highlevel
ones (data: EX GDd vectors, code: Scheme and PF interpreters).
This is about the total elimination of _all_ Scheme and PF code in a
final project, to get rid of the management requirement of not _using_
any unknown or non-standard languages, so the resulting C code can be
modified without needing any access to the higher level data and code