Path: utzoo!attcan!utgpu!news-server.csri.toronto.edu!mailrus!sharkey!sevihc!reso
From: sevihc!reso@sharkey.cc.umich.edu (Dennis Reso)
Newsgroups: comp.lang.c
Subject: C portability
Message-ID: <1990Aug8.030538.16906@sevihc.UUCP@sharkey.cc.umich.edu>
Date: 8 Aug 90 03:05:38 GMT
Distribution: usa
Organization: Total Lack Thereof
Lines: 29

Where I currently work, we have a lot of FORTRAN programmers and
are beginning to use C in several areas only because FORTRAN just
*can't* do what we need.  The problem is, we need to specify a 
set of recommendations and checkers to make sure (as best as we
can) that code remains portable to several (SYSV, BSD, PrimeOS)
machines. We currently have a checker for the FORTRAN, but my
experience with C indicates that we are much better off educating
programmers rather than spending a lot of time second-guessing
and correcting them. Therefore I decided to tap into the collective
net-wisdom concerning:

1. Defensive programming.
   Are there books available which discuss, in depth, programming
   practices that result in the most machine-portable code?

2. Lint as a portability checker.
   Is there a description of what lint is supposed to do, for
   example sufficient to write your own if a machine has none.
   (Not that we're planning on it, but Prime has no lint and it
   would be nice to know that there is some kind of "standard"
   regarding lint).
   Since it seems to be integrated with the compiler and libraries,
   how machine-specific are lint runs on the same code/different 
   machines likely to be?

________________________________________________________________________
Dennis Reso                        sevihc!reso@sharkey.cc.umich.edu
Sterling Hieghts, MI USA           {sharkey|itivax}!sevihc!reso
Ford Motor Company, Dearborn       pms415!reso@fmsrl7.srl.ford.com