Path: utzoo!attcan!uunet!know!zaphod.mps.ohio-state.edu!ncar!gatech!rutgers!rochester!pt.cs.cmu.edu!dsl.pitt.edu!pitt!willett!ForthNet
From: ForthNet@willett.pgh.pa.us (ForthNet articles from GEnie)
Newsgroups: comp.lang.forth
Subject: Metacompilation
Message-ID: <1890.UUL1.3#5129@willett.pgh.pa.us>
Date: 21 Oct 90 01:58:46 GMT
Organization: String, Scotch tape, and Paperclips.  (in Pgh, PA)
Lines: 47


   To: SHANNON VANCE                 Refer#: 1776
 From: RAY DUNCAN                      Read: NO
 Subj: METACOMPILER                  Status: PUBLIC MESSAGE
 1.  Have you tried the file ROMFORGE.ZIP in conference 5?  As far as I
 know this version of Forget is "good."
 2.  The metacompiler doesn't do any magic hidden things at the end of
 the compile.  All changes to the image are explicit and are in the
 target's source file (e.g. HERE INIT-DP ! and so on).  You just need to
 calculate the checksum AFTER these interpreted actions at the end of the
 compilation.  Or am I missing something here?
 3.  This is reasonable and we plan to do it.  However there are some
 funny implications by the fact that the Metacompiler makes 2 passes and
 forward references are allowed.  You might not get exactly what you
 expect if you aren't careful to avoid forward references.
 4.  A nonzero return code, by convention, means an error has occurred. 
 This is pretty standard behavior for compilers and other utilities.  How
 could we preserve this (thereby allowing people (including ourselves) to
 use the Metacompiler with a MAKE utility) and still give you the ability
 to set ANY return code you want at the end of compilation?
 5.  The metacompiler currently puts the output files wherever it found
 the input files.  It seems reasonable to me to specify a different path
 for the output; I'll look at this.
 6.  Using an environment variable as part of a path sounds reasonable to
 me, and shouldn't be too hard to add.
 7.  I'm curious about what sort of DOS command you would want to spawn
 during a metacompile (which is not an interactive process).  There is
 very little memory around when the metacompiler is running; I doubt if
 there is enough to run any problem of reasonable size.  We use just
 about all of it for symbol tables & stuff that we don't want to page in
 and out from disk (the only thing that we page is the image file).
 8.  Testing environment variables for .IF statements should be possible.
 But I'm not sure I'm ready to commit to this sort of thing.  This would
 lead us into trying to make all the string operators "interpretable" at
 metacompile time and has other hidden ramifications.
 32.  If you are using control-Break to terminate a metacompile run, I
 can't tell you anything except the behavior of Control-Break varies from
 one ROM BIOS to another.  It works ok on my machines, and we have
 special code in there to handle various ROM BIOS peculiarities, but we
 can't check all possible machines.  Perhaps we'll just have to disable
 the control-break capability altogether during metacompilation and
 monitor the keyboard for entry of some other character (e.g. Control-C).

 NET/Mail : LMI Forth Board, Los Angeles, CA (213) 306-3530             
-----
This message came from GEnie via willett through a semi-automated process.
Report problems to: dwp@willett.pgh.pa.us or uunet!willett!dwp