From: andy AT netfall DOT com
Message-Id: <200001102348.PAA2671404@meer.meer.net>
Date: Mon, 10 Jan 2000 14:29:28 -0800 (PST)
Subject: Re: Possible bug
To: pgcc AT delorie DOT com
In-Reply-To: <3879F657.93592E00@altmayer.com>
MIME-Version: 1.0
Content-Type: TEXT/plain; CHARSET=US-ASCII
Reply-To: pgcc AT delorie DOT com
Errors-To: dj-admin AT delorie DOT com
X-Mailing-List: pgcc AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk

On 10 Jan, Juan Altmayer Pizzorno wrote:
> Hi,
> 
> the attached file was "factored out" of a program that, among other
> things, does DNS queries.  The code between the printf()'s seems to
> affect the value of variable 'p', even though it shouldn't...  When
> compiled with -O0 or no -O flags, it runs correctly.  But when compiled
> with -O1 and above, it does not.  I've tried it with several different
> versions of the compiler (egcs 2.91.66 from plain RedHat, egcs 2.95.2
> from the official release, and pgcc 2.95.3 with the pgcc-2_95_3 tag
> out of your CVS), all with identical results.  BTW pgcc-2_95_3 still
> identifies itself as 2.95.2...  The second attachment is "t.E", obtained
> with g++ -E t.cxx > t.E.  This is all on a Pentium II, RedHat 6.1,
> kernel 2.2.14-based notebook.  Thanks

Geez, at least next time could you run the preprocessor output through
uniq?

a