From hirchert@ncsa.uiuc.edu Fri Mar 12 06:05:07 1993
Received: from newton.ncsa.uiuc.edu by dkuug.dk with SMTP id AA05034
  (5.65c8/IDA-1.4.4j for <sc22wg5@dkuug.dk>); Fri, 12 Mar 1993 18:59:43 +0100
Received: from troon20.ncsa.uiuc.edu by newton.ncsa.uiuc.edu with SMTP id AA17872
  (5.65a/IDA-1.4.2 for sc22wg5@dkuug.dk); Fri, 12 Mar 93 11:59:58 -0600
Return-Path: <hirchert@ncsa.uiuc.edu>
Received: by troon.ncsa.uiuc.edu (4.1/NCSA-4.1)
	id AA21345; Fri, 12 Mar 93 12:05:10 CST
Message-Id: <9303121805.AA21345@troon.ncsa.uiuc.edu>
From: hirchert@ncsa.uiuc.edu (Kurt W. Hirchert)
Date: Fri, 12 Mar 1993 12:05:07 CST
X-Mailer: Mail User's Shell (7.2.1 12/20/90)
To: jtm@llnl.gov
Subject: N865 ballot
Cc: sc22wg5@dkuug.dk
X-Charset: ASCII
X-Char-Esc: 29

I vote YES on all items except 39 (which was withdrawn and thus gets
no vote whatsoever) and 13, on which my vote is YES WITH the COMMENTS
below:

In reading the comments accompanying many of the negative votes on
this item, I find what I believe to be misconceptions about it.  I
would like to set the record straight:

  The subcommittee that prepared this item believed that there is
  an inconsistency in the standard in that two features that were
  intended to behave the same way (i.e., the accessibility of
  entities in a nested scoping unit and the accessibility of
  type mapping rules in a nested scoping unit) were made to
  behave differently by a late change that was made to one and
  not the other.  The subcommittee believed that it was intended
  that the action to apply to both features and that the
  failure to apply it to one was an inadvertent editorial error.
  Thus, the intent of this item is _not_ to change previously
  agreed upon technical content, but rather to make the text
  of the standard reflect the content that was agreed upon.

I would also like to comment on a couple of specific points:

o John Reid and Lawrie Schonfelder seem to disagree on the
  chronology of this text, with Lawrie saying it was intentionally
  changed late in the process and John saying it was adopted
  in March 1987 and unchanged for more than 4 years after that.
  [Given the overall length of the process, I suppose it
  could be that Lawrie perceives four years before the end to be
  late in the process.]  In any case, the key chronology from the
  perspective of the subcommittee is that the change to isolate
  interface bodies from the surrounding scoping unit was made
  _after_ the last time this text was updated.

o Lawrie presents a strong example of why he thinks implicit
  rules (especially IMPLICIT NONE) should propagate.  I personally
  agree with this argument and many of the other arguments why
  interface bodies should not have been isolated from their
  hosts, but Lawrie and I lost that argument when X3J3 voted
  to isolate interface bodies.  As a responsible maintainer, I
  do not feel it appropriate to get my own way through a back door
  editorial accident, and so I am voting to make the text reflect
  X3J3's decision at the time is was developing the standard, not
  my personal position (then or now).

To those that contend that the current text is not inconsistent,
I ask a question:  Of what value is isolating interface blocks
from host association without also isolating it from host
implicit rules?  I have been unable to construct even one
believable example that benefits from the late change to host
association in isolation.  (I _can_ construct examples that
are affected by the change, but all contain artifacts that I
have no reason to expect to find in real programs.)

-- 
Kurt W. Hirchert     hirchert@ncsa.uiuc.edu
National Center for Supercomputing Applications