From drlevine@apollo.hp.com Fri Apr 15 13:39:34 1994 Received: from amway.ch.apollo.hp.com by dkuug.dk with SMTP id AA19255 (5.65c8/IDA-1.4.4j for ); Fri, 15 Apr 1994 23:44:55 +0200 Message-Id: <199404152144.AA19255@dkuug.dk> Received: from hillside.ch.apollo.hp.com by amway.ch.apollo.hp.com id Fri, 15 Apr 94 17:39:36 EDT Received: by hillside.ch.apollo.hp.com id ; Fri, 15 Apr 1994 17:39:35 -0400 From: drlevine@apollo.hp.com Date: Fri, 15 Apr 94 17:39:34 EDT To: martin@ocfmail.ocf.llnl.gov (Jeanne T Martin) Cc: sc22wg5@dkuug.dk X-Charset: ASCII X-Char-Esc: 29 Subject: Corrigendum 2 letter ballot response I'd like to submit the following vote on the Corrigendum 2 ballot. David Levine Hewlett-Packard Company HP Languages - Chelmsford I vote YES without comment, on all items, with the following exceptions: 00004 YES The committee's position is clear and unambiguous. My personal taste tends more towards the position enunciated by Lawrie Schonfelder: that these should be considered as multi-character tokens and written without internal spaces. 00100 NO Hendrickson, Moss and Schonfelder have presented some compelling arguments in their recent comments on this issue. I'd like to join their negative vote. Clearly, there are implementation tradeoffs involved in the support of zero-sized objects. The current proposal does not represent an adequate solution for a Standard; rather, it allows the behavior to become Implementation Defined. That does not further the vital aims of program portability and predictable behavior. Given the need for a pragmatic compromise, the proper choice appears to allow all zero-sized objects (of the same shape, etc.) to be truly associated. 00129 YES Insofar as this response merely reinforces the existing language of the Standard, there's no reason not to let it proceed. I agree with Moss' comment, that a change would be appropriate to correct the problem. The standard produces an unnecessary anomaly in the language, distinguishing between intrinsic operations and intrinsic functions in a way which is not done anywhere else. It is true that the current text, from a strictly legalistic interpretation, does support this restriction. I find it just as easy to argue that the text presents a recursive concept, but that through a failure in precision of language has overlooked one of the ways in which an implied-DO variable could appear. (Note that the restrictionin (4) is to an "initialization expression", not a "constant" as in (1) where the full generality is explicitly excluded. Even the bounds in (2) are allowed to be implied-DO variables.) David Levine HP Languages - Chelmsford -------