Fri Aug 29 16:19:24 CEST 2008

two distinct uses of byte code

I'm sick, so this might not make sense.

There are 2 uses of byte code:

  * as decoupling interface
  * as serialization protocol

In the first case, bit patterns are associated to specific semantics.
All programming uses this byte code explicitly, possibly from a shared
table where it's specified only once, used by both sides of the

In the latter case, byte code is a consequence of having to squeeze
information over a channel between two componets of the SAME system,
and its values are essentially arbitrary: if they are never touched by
a programmer, they can be generated automatically.  This allows for
COMPRESSED byte codes: ones that result in dense code.

The are destinguished by the code being published or not.