Fri May 9 18:35:17 CEST 2008

assembler expression language

what currently is 'target:' might better be written in s-expr syntax,
so that it's easily converted to concatenative syntax. (the other way
is more difficult).

while now it's kind of cute to have this concatenative language map to
a concatenative assembler expression language, later when external
assemblers need to be supported, this might become a nuisance.

also, and probably more important: a distinction needs to be made
about data types and partial evaluation:

  * use scheme's infinite precision types
  * use only target types (i.e. accurately SIMULATE the computation)