Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!unmvax!deimos.cis.ksu.edu!ux1.cso.uiuc.edu!ux1.cso.uiuc.edu!uxa.cso.uiuc.edu!ews00461
From: ews00461@uxa.cso.uiuc.edu
Newsgroups: comp.sys.mac.programmer
Subject: CODE segments
Message-ID: <227700050@uxa.cso.uiuc.edu>
Date: 10 Nov 89 10:32:24 GMT
Lines: 23
Nf-ID: #N:uxa.cso.uiuc.edu:227700050:000:1055
Nf-From: uxa.cso.uiuc.edu!ews00461    Nov  9 10:27:00 1989



I find the discussion of segmentation very interesting, and I have a
few questions which are probably more basic...

Suppose I was going to write a compiler for the Mac.  My understanding of
what this beast should finally create is this:
	The output would perhaps be a file of type APPL with a number
of CODE segments (among a few other things).  The CODE segments themselves
are of interest to me.  Such CODE segments should contain completely
relocateable code.  All global variables should be reference from A5,
(where does A5 get set?).  All local variables should be on the stack.
All jumps should be relative, if they are in the same segment, and
should go through the jump table if they are to another segment (how is
this done, I assume there is a toolbox call to do this?)  There must
be at least 2 (?) segments, 1 jump and 1 with real code.

Any insight into this whole area would be very interesting.  And
obviously, there are a few people on this net for whom this is
trivial.  Any wisdom to pass on ?

Eric W Sink
ews00461@uxa.cso.uiuc.edu