ISO/IEC JTC1/SC22/WG14 N578


Proposal UK006b - Restructuring of subclauses 6.1 and 6.2
=========================================================

Summary
-------
Subclauses 6.1 and 6.2 are badly organised, and the former contains a
number of ambiguities and inconsistencies. This proposal is a general
cleanup.


Conformance
-----------
The author believes that no strictly-conforming program is affected by this
proposal, and that the only changes are in the area of ambiguities where all
existing implementations make the "right" choice.


Discussion
----------
The following changes have been made.

(1) The important concepts of scope, linkage, namespaces, storage durations,
and types have been separated out from a subclause of a lexical concept to
a major element of their own. To minimize impact, the text on conversions
(6.2) is merged with this to become the new 6.1. The remainder of 6.1,
which is actually the lexical elements, becomes 6.2. This aids the reader
of the standard and reduces the need for forward references. It does not
alter the language defined.

(2) The current standard attempts to treat both operators and punctuators
as syntactic elements. This means that several tokens (including the
keyword "sizeof" occur twice. Instead, punctuator is made a syntactic
concept, and operator a semantic one. The effect is to eliminate some
minor syntactic ambiguities (various tokens have multiple derivations) and
make the concepts clearer. It does not alter the language defined.

(3) The header name changes of TC1 have been further amended for clarity.
It does not alter the language defined.

(4) Various nugatory constraints are eliminated. Any program which violates
these constraints also violates one or more syntax rules. It does not alter
the language defined.

(5) The semantics of integer character constants are clarified. The
sentence which refers to footnote 25 has been removed, and footnotes 24 and
25 combined and reworded.


Detailed proposal
-----------------
This proposal is best appreciated by reading the resulting text as a whole,
rather than examining the detailed changes. The former is available for
examination at the URL [not available publically].

This text is generated by the following changes, which form the proposal.

Restructure subclauses 6.1 and 6.2 according to the following table, and
alter cross-references throughout the Standard accordingly.

  New number   Title                                Old number
  ------------------------------------------------------------
  6.1          Concepts                             none
  6.1.1        Scopes of identifiers                6.1.2.1
  6.1.2        Linkages of identifiers              6.1.2.2
  6.1.3        Name spaces of identifiers           6.1.2.3
  6.1.4        Storage durations of objects         6.1.2.4
  6.1.5        Types                                6.1.2.5
  6.1.6        Compatible type and composite type   6.1.2.6
  6.1.7        Conversions                          6.2
  6.1.7.1      Characters and integers              6.2.1.1
  6.1.7.2      Signed and unsigned integers         6.2.1.2
  6.1.7.3      Floating and integral                6.2.1.3
  6.1.7.4      Floating types                       6.2.1.4
  6.1.7.5      Usual arithmetic conversions         6.2.1.5
  6.1.7.6      Lvalues and function designators     6.2.2.1
  6.1.7.7      void                                 6.2.2.2
  6.1.7.8      Pointers                             6.2.2.3
  6.2          Lexical elements                     6.1
  6.2.1        Keywords                             6.1.1
  6.2.2        Identifiers                          6.1.2
  6.2.3        Constants                            6.1.3
  6.2.3.1      Floating constants                   6.1.3.1
  6.2.3.2      Integer constants                    6.1.3.2
  6.2.3.3      Enumeration constants                6.1.3.3
  6.2.3.4      Character constants                  6.1.3.4
  6.2.4        String literals                      6.1.4
  6.2.5        Punctuators                          none
  6.2.6        Header names                         6.1.7
  6.2.7        Preprocessing numbers                6.1.8
  6.2.8        Comments                             6.1.9

Delete the old subclauses 6.1.5 and 6.1.6; the new subclause 6.2.5 is given
below.


Insert and remove forward references according to the following table.

[never completed]


======

In subclause 6.2 (formerly 6.1) syntax, remove the alternative "operator"
from both "token" and "preprocessing-token". In the constraints, remove
"an operator,". In the semantics, remove "operators," (twice), and combine
the last two paragraphs into one, replacing:

    A header name

at the start of the last paragraph, with

  There is one exception to this rule: a header name


In subclause 6.2.2 (formerly 6.1.2), delete the constraints. In the semantics,
replace:

  that will be described later:

with

  described in subclause 6.1.1:

Move the last paragraph of the implementation limits to the end of the
semantics.


In subclause 6.2.3.4 (formerly 6.1.3.4) description, delete the sentence:

  If any other escape sequence is encountered, the behaviour is
  undefined.25

Delete footnote 25. Add the following text at the end of footnote 24:

  If any other character follows a backslash, the result is not a
  token and a diagnostic is required. See ``future language
  directions'' (6.9.2).

Replace the first paragraph of the semantics with:

       An integer character constant has type int.  The  value
  of   an  integer  character  constant  containing  a  single
  character that maps into a member  of  the  basic  execution
  character  set  is the numerical value of the representation
| of that character converted to type char and then to type int.
| The value of an integer character constant containing a single
| octal or hexadecimal escape sequence shall be the value of
| that sequence converted to type char and then to type int. The
  value  of an integer character constant containing more than
  one character, or containing a character or escape  sequence
  not  represented  in  the  basic execution character set, is
  implementation-defined.

(only the four marked lines are actually changed).


Insert a new subclause 6.2.5 (replacing the old 6.1.5 and 6.1.6):


  Syntax

          punctuator: one of
                  [  ]  (  )  {  }  .  ->
                  ++  --  &  *  +  -  ~  !
                  /  %  <<  >>  <  >  <=  >=  ==  !=  ^  |  &&  ||
                  ?  :  ;  ...
                  =  *=  /=  %=  +=  -=  <<=  >>=  &=  ^=  |=
                  ,  #  ##  <:  :>  <%  %>  %:  %:%:

  Semantics

       A punctuator is a symbol that has independent syntactic
  and semantic significance. Depending on context, it may specify
  an operation to be performed (which in turn may yield a value
  or a designator, or produce a side effect, or some combination)
  in which case it is known as an *operator* (other forms of
  operator also exist in some contexts). An *operand* is an entity
  on which an operator acts.

       In all aspects of the language, these six tokens

          <:  :>  <%  %>  %:  %:%:

       behave, respectively, the same as these six tokens

          [   ]   {   }   #   ##

       except for their spelling.27


  Forward references:  expressions (6.3), declarations  (6.5),
  preprocessing directives (6.8), statements (6.6).

  __________

  27. Thus [ and <:  behave  differently  when  ``stringized''
      (see  subclause  6.8.3.2),  but  can otherwise be freely
      interchanged.
  __________


In subclause 6.3.17, replace the first sentence of the example with:

       As indicated by the syntax, the comma operator (as described in
  this subclause) cannot appear in contexts where a comma is used to
  separate items in a list (such as arguments to functions or lists
  of initializers).