Path: utzoo!utgpu!watserv1!watmath!att!dptg!ulysses!andante!princeton!udel!wuarchive!usc!elroy.jpl.nasa.gov!lll-winken!diego.llnl.gov!miller
From: miller@diego.llnl.gov (Patrick Miller)
Newsgroups: comp.lang.functional
Subject: Error Values (Revisited)
Keywords: functional error implementation compiler
Message-ID: <86826@lll-winken.LLNL.GOV>
Date: 27 Nov 90 17:32:07 GMT
Sender: usenet@lll-winken.LLNL.GOV
Reply-To: miller@diego.llnl.gov (Patrick Miller)
Organization: Lawrence Livermore National Laboratory
Lines: 18
Nntp-Posting-Host: diego.llnl.gov

I didn't realize that I would spark such discussion!  Anyway, thanks
to all who responded.  More responses would still be appreciated
though.

In my original posting, I wasn't very specific enough about what
constituted an implementation.  As such, some of the replies that I
received were about tackling the problem of specifying the semantics
of ERROR values.  These replies were useful and interesting.   I
would like to propose a more specific request -- Given a set of
error semantics, how does one effectively COMPILE the code?

-- 
Pat Miller				     Professional Student
miller@lll-crg.llnl.gov			       Amateur Politician

"If I knew what I was doing, I wouldn't be running"

Fight Apathy!				   Vote Early, Vote Often