Path: utzoo!attcan!uunet!ubvax!ardent!rap
From: rap@ardent.UUCP (Rob Peck)
Newsgroups: comp.sys.amiga
Subject: Re: Ideas for Next Arp
Message-ID: <732@ardent.UUCP>
Date: 23 Nov 88 18:30:00 GMT
References: <8811221116.AA23400@jade.berkeley.edu>
Organization: Dana Computer, Inc., Sunnyvale, CA
Lines: 25

In article <8811221116.AA23400@jade.berkeley.edu>, U211344@HNYKUN11.BITNET (Olaf 'Rhialto' Seibert) writes:
> I have an (exellent) idea for the next release of Arp:
> 
> Put the functionality of REZ into the RESISIDENT command of ARP.
> 
> Since the 'principles of operation' of Rez are contained in its
> documentation, it should not be too difficult to duplicate.

As I understood it from Jim Goodnow [some roman numeral], who wrote REZ,
the primary reason it works is that he recognizes some special sequence
of code that is in the Manx 3.6 (and probably later) distributions.  To
make a REZ-compatible system work for ALL compilers would entail some
cooperation on the part of the compiler writers, or some way of recognizing
for each level of each compiler, which compiler produced the code and
an appropriate way of linking it.  If you take a real good look at the
theory of ops Jim has provided, you'll see that it might not be quite
so easy to adapt it to a program done by "any" compiler.  For example,
(as I recall), if there are program globals, he allocates a separate
area of memory for the globals for each new incantation of the program.
Unless you can properly identify all of the hunks, this becomes a
problem.  Jim's implementation of REZ, though, means that a program
need not be designed to be re-entrant in the first place, because each
new incantation looks like a fresh load of the program.

Rob Peck