Hi,
I really didn't think of this befoere. Maybe what Said() does is nothing
(...) more than generating and evaluating a LISP expression from the
said spec (which then is nothing more than a simplified notation). I
mean, try "strings sciv.exe" - hmmm... "augment_tree err: non-consp
KID". Smell some LISP yet? :-)
(OK, this may be really far-fetched, I just dug out an elisp reference
from the 'net - I'm just learning) Each node in the parse tree would
then be a cons cell (with the left pointer beiing the car pointer and
the right pointer being the cdr pointer), and augment_tree is really
just an implementation of the cons function. The storage codes are used
to identify the type of the subordinated list or atom. With that in mind
it would
probably be easier to write a full Said() implementation.
Another thing:
What are we going to do about patented algorithms? Once we get to SCI01,
we'll have the path avoidance system to take care of (which is
documented in patent number 5287446 available at patents.ibm.com). Also,
the PKWARE compression used in later SCI1's may cause us problems.
That's all for now,
Lars
