Path: utzoo!censor!geac!torsqnt!news-server.csri.toronto.edu!cs.utexas.edu!usc!zaphod.mps.ohio-state.edu!uakari.primate.wisc.edu!aplcen!haven!adm!smoke!gwyn
From: gwyn@smoke.brl.mil (Doug Gwyn)
Newsgroups: comp.lang.c
Subject: Re: Re: How ANSI is TC++?
Message-ID: <14374@smoke.brl.mil>
Date: 7 Nov 90 10:45:24 GMT
References: <5940044@hpcupt1.cup.hp.com> <1990Oct26.162147.8767@ux1.cso.uiuc.edu> <131@nazgul.UUCP>
Organization: U.S. Army Ballistic Research Laboratory, APG, MD.
Lines: 20

In article <131@nazgul.UUCP> bright@nazgul.UUCP (Walter Bright) writes:
>A lot of people talk about "100% ANSI C compliance". This is impossible,
>as it implies there are *no* bugs in the compiler. We all know this
>is unattainable (by any known technology!). The best one can say is
>it passes so-and-so's test suite, or was validated by such-and-such
>outfit.

I think you might get an argument from the folks at Metaware, among others.
The technology does exist.  Indeed, the weak point right now seems to be
the fact that the standard is written in English instead of a formal
semantic specification language.  (That was an intentional decision, by the
way, in an attempt to ensure that the standard would be useful to a wide
audience and not just a few specialists in the technology.)

It is true, however, that practically all implementations have not been
proven conformant.  If the implementors of a compiler are not aware of
any shortcomings, and if their product has been subjected to thorough
testing (including as many conformance test suites as possible), then it
would be reasonable to call it "a conforming (hosted/freestanding)
implementation" until such time as proven otherwise.