Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site mruxe.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!bellcore!petrus!sabre!zeta!epsilon!gamma!pyuxww!mruxc!mruxe!cgw
From: cgw@mruxe.UUCP (C Waldman)
Newsgroups: net.arch
Subject: linking
Message-ID: <105@mruxe.UUCP>
Date: Sat, 24-May-86 14:42:51 EDT
Article-I.D.: mruxe.105
Posted: Sat May 24 14:42:51 1986
Date-Received: Sun, 25-May-86 18:41:59 EDT
Organization: Bell Communications Research, Piscataway, NJ
Lines: 22



    The recent discussion about linking time has inspired me to start a
project (admittedly not finished yet) which I am implementing on an IBM
PC, but the idea seems like it could be easily implemented on any system.
It is basically a testing linker, not for producing production code but
simply to speed up the test/debug/recompile/relink cycle that drives me
up the wall. I am using Lattice version 3,  with no complaints, but the
Microsoft linker leaves a lot to be desired.  The prototyping linker I'm
talking about is simply an interpreter (OK, it's slow) which loads a bunch
of object modules, begins executing at main, and resolves external        
references at run-time.  Then when everything checks out OK, I can use 
LINK to produce an .EXE file. 
    Most of you probably have a lot more experience than me with this
kind of thing, so I wold appreciate any feedback concerning why this
is/is not a good idea.


    ...{ihnp4|bellcore}!mruxe!cgw
    Charles G. Waldman, Bell Communications Research, Morristown, NJ

,p