Thu Aug 27 15:03:09 CEST 2009

simpler primitives

As mentioned before: there are too many things to think about
implementing linear primitives (possible GC restart and mutation).
This reduces my confidence that the code is correct in the presence of
primitive error aborts or GC aborts.

So, either I give up on the idea of GC restarts, or I factor the code
such that these distinctions are no longer a problem.