File:  [gforth] / gforth / model
Revision (vendor branch): download - view: text, annotated - select for diffs
Fri Feb 11 16:30:47 1994 UTC (29 years, 10 months ago) by anton
Branches: no-vendor
CVS tags: alpha
The GNU Forth Project

$Id: model,v 1994/02/11 16:30:47 anton Exp $
Copyright 1992 by the ANSI figForth Development Group

This file describes the implementation model of ANSI figForth. The things
described herein are valid in all implementations of ANSI figForth.

The instruction pointer (ip) is post-incremented.

Branches are relative:
offset:	target - offset
target:	...

The top-of return stack is in memory

The top of data stack need not be in memory. Other data stack items
are in memory.

The stacks grow downward

flags are represtented by 0 and -1

The loop control parameters are the index on the top of return stack
and the limit on the second item.

Everything except the virtual machine registers (ip, sp, etc.) is

The return address on the return stack occupies one cell.

There is a floating point stack
The floating point numbers are the same as C's double numbers

The error returns of the file word set are FALSE or TRUE. (or should
we use errno? Does this work with stdio functions?)

No assumptions can be made about the order of fetching the xt of the
next word wrt the other operations in a forth word. This makes a
difference only if you want to execute code lying on the stack or if
you store into the very next word that is executed.

FreeBSD-CVSweb <>