Xref: utzoo comp.lang.c:11573 comp.arch:5801
Path: utzoo!utgpu!water!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!mailrus!cornell!uw-beaver!teknowledge-vaxc!sri-unix!garth!smryan
From: smryan@garth.UUCP (Steven Ryan)
Newsgroups: comp.lang.c,comp.arch
Subject: Re: compiler detection of potential run-time errors
Message-ID: <1103@garth.UUCP>
Date: 29 Jul 88 20:29:44 GMT
References: <1988Jul23.221743.22169@utzoo.uucp> <477@m3.mfci.UUCP> <1075@garth.UUCP> <1988Jul27.202049.21589@utzoo.uucp>
Reply-To: smryan@garth.UUCP (Steven Ryan)
Organization: INTERGRAPH (APD) -- Palo Alto, CA
Lines: 12

>My standing comment on this one is "programmers do not write arbitrary
>programs".  I have no objection to monkey programmers being told "this
>program is so poorly written that verifying the absence of run-time errors
>is beyond this compiler's ability".  In fact, I have no objection to real
>programmers occasionally getting the same message, provided some attempt
>is made to identify the source of the problem.

A point to remember is the compiler doesn't know who or what wrote the code.

I take it then the idea to do prove what can be proved, refute what can be
refuted, and it's the middle ground which is rejected. I see no objections
to this approach if it is well defined.