Dan, Jarkko, etc.  How about we try to identify the big units we'll be
working on?

  parser
  optimizer
  bytecode
  runtime dispatcher (w/knowledge of dynaloading)
  data-types
  regex
  memory

My idea of development is something like this:

 * identify the units/modules
 * map relationships between modules
 * design the API for the modules
 * code test cases for the API (circular process with the previous
   step?)
 * code the API
 * code the wrappers for the libraries (perl.c, etc)

Make sense?

Nat

Reply via email to