From maine@altair.dfrf.nasa.gov Wed Jan 11 15:49:05 1995 Received: from altair.dfrf.nasa.gov by dkuug.dk with SMTP id AA16961 (5.65c8/IDA-1.4.4j for ); Wed, 11 Jan 1995 17:49:41 +0100 Received: by altair.dfrf.nasa.gov (5.0/SMI-SVR4) id AA12482; Wed, 11 Jan 1995 07:49:05 +0800 Date: Wed, 11 Jan 1995 07:49:05 +0800 From: maine@altair.dfrf.nasa.gov (Richard Maine) Message-Id: <9501111549.AA12482@altair.dfrf.nasa.gov> To: SC22WG5@dkuug.dk In-Reply-To: <199501111326.AA09598@dkuug.dk> (message from Dick Hendrickson on 11 Jan 1995 07:26:14 -0500 (CDT)) Subject: Re: (SC22WG5.674) Re: informal ballot of 17 defect responces Content-Length: 0 X-Charset: ASCII X-Char-Esc: 29 On 11 Jan 1995 07:26:14 -0500 (CDT), Dick Hendrickson said: Dick> The answer is that there isn't a good answer. I definitely agree with that. Dick> We had to pick something... But I disagree with that conclusion. If there really isn't a good answer, then I'd rather see it made processor-dependent than explicitly forced to something that might be "right" for some cases but is "clearly wrong" for others. By decreeing a particular answer, you are forcing the processor to put in a special case check for the zero-sized case just in order to get it "wrong." And what are the odds that some processors won't have the check even if the standard (in its third corrigendum) technically requires it, so that in practice the users will still need to treat the result as processor-dependent? (Just like a user had better consider the result of backspacing over an end-of-file record as processor dependent because too many processors don't really follow the standard). I'd really like to see the "base address approach", but Dick is right that the standard doesn't define this and it would take a lot or work to make it unambiguously defined for all cases. From my user perspective, I'd hate to have to explain this special quirk to other users. It is much easier to tell them that association of zero-sized objects is ill-defined and shouldn't be counted on than to tell them to remember the special rule that all such associations are (.true. or .false. - pick one). Either choice is going to seem completely arbitrary and counter-intuitive in some cases. -- Richard Maine maine@altair.dfrf.nasa.gov