[<<][libprim][>>][..]
Fri Nov 27 11:36:23 CET 2009

remove support for towers

It makes the interpreter loop too complicated, and it's not used.

This means sc_step is not a function accessible from scheme, which
means it can be implemented more efficiently, and probably simpler.

The fundamental conflict is this:  some primitives need explicit
access to the current interpreter state - this conflicts with
implementing the interpreter in a purely functional way.

Essentially, the concept is broken so let's get rid of it.  The vm is
a state-machine because of a hack that allows modification of internal
state (i.e. the current value & continuation) in an interface towards
primitives that pretends as if there is no suck access.

Is there a reason to not unify primitive exceptions and GC restarts?



[Reply][About]
[<<][libprim][>>][..]