07-199
To: J3
From: Malcolm Cohen
Subject: UTI 111
Date: 2007 May 01

1. Introduction

This paper addresses UTI 111, in conjunction with the Terms and definitions
paper which has not yet been written.

2. Notation

As an experiment I am going to be using

  <ref> <<is>> [ <page-no> : <subclause-no> p <para-no> ]
        <<or>> [ <page-no> : <subclause-no> p <para-no> + <line offset> ]
        <<or>> [ <page-no> : Note <number> ]
        <<or>> [ <page-no> : UTI <number> ]

to refer to the standard by paragraphs instead of page/line.
For example, [461:15.2.3.1p2+3] is on page 461 and is the fourth line of
paragraph 2 of subclause 15.2.3.1.

(This is a bit more verbose than our usual one: for a start, the page
number is redundant, but it makes it faster to find things on paper.)

3. Discussion

(a) There are some edits to do with notes here, which have nothing to do
    with UTI 111 apart from being in fairly close proximity.

(b) The term "address" I believe actually comes from ISO/IEC 2382-1:1993,
    Information technology -- Vocabulary -- Part 1: Fundamental terms.
    The C standard itself does not define the term!  Therefore we should
    almost certainly reference the vocabulary as well.

4. Edits to 07-007r1

[460:15.2.2 Note 15.2]
  Delete this nonsensical meaningless note.
{A "companion processor" cannot "support more than one variety of float",
 because by the C standard itself there is only one variety of float!
 Under the meaning of the act, a C compiler that uses command-line switches
 to control the meaning of "float" is providing more than one C
 implementation or companion processor.

 In any case, the need for some means of choosing which companion processor
 (whether expressed in a single program via options or via multiply
 programs) is in no sense limited to float, double and long double.
 Therefore this note is not just meaningless and incorrect, it is also
 misleading!

 And in no sense would it justify breaking up the subclause by plonking it
 in the middle either (the ISO guidelines say notes go at the end of the
 subclause unless there is good reason), even were it not incorrect and
 misleading.}

[460:15.2.2 Note 15.3]
  Move this note to the end of the subclause (4 lines down).
{Editorial.}

[461:15.2.3.1p2]
  Delete unnecessary and misplaced paragraph.
{It contains lots of unnecessary witter, as mentioned in the UTI, and what
 little content it does have is faulty.}

[461: UTI 111]
  Delete UTI when satisfied it has been resolved.

Add the following definition to the terms and definitions clause.
  "2.5.10.x
   <<C address>>
   the address of an object either defined by the companion processor or
   which might be accessible to the companion processor; this is the same
   concept which the C International Standard calls the address.

   [ISO/IEC 2382-1:1993]"
{The placement is deliberately vague because our terms and definitions
 clause does not even come close to adhering to the ISO rules.  Anyway,
 2.5.10 is where we talk about companion processors right now, so in the
 event of my not being ready to produce a proper terms and definitions
 clause I can turn the above into a normal sentence and stuff it in there.
 An alternative would be to put in into 15.1 as a third paragraph

 BTW, the text mentions the C standard because it's true and also serves to
 motivate the name, the reference at the bottom references the vocabulary
 standard because that's where the term *actually* comes from.  I think.}

[7:1.9 all paragraphs but the first]
  Sort into ISO number order and add
    "ISO/IEC 2382-1:1993, Information technology -- Vocabulary --
     Part 1: Fundamental terms."
{We are supposed to reference the standard that defines a term, not just
 a random other standard that uses it.  There is no particularly useful
 order (importance?!) for the references, so we should just list them
 in numeric order.}

===END===