Re: Range of enums

[Mark Mitchell <mark at markmitchell dot com>]

>>>>> "Jeffrey" == Jeffrey A Law <law@hurl.cygnus.com> writes:

    Jeffrey>   In message <14020.22010.51278.609918@janus.pgt.com>you
    Jeffrey> write:
    >> I think this is a not-too-uncommon idiom in C to step through
    >> all the values of an enum.  As far as I know this is legal C;
    >> I'd really like to know if it isn't.

    Jeffrey> That's precisely what we're trying to determine.  Whether
    Jeffrey> or not the code is valid has a significant effect on how
    Jeffrey> we approach the problem.

My reading of the C9X draft is that the construct in question *is*
legal, provided the smallest integer type is at 3 bits:

  6.5.2.2:
  
    Each enumerated type shall be compatible with an integer type.
    The choice of type is implementation-defined, but shall be capable
    of representing the values of all the members of the enumeration.

  6.1.2.6:

    Two types are "compatible" if their types are the same.

If other words, each enumerated type is "the same" as an integer
type.  Since our smallest integer type has 8 bits, the example posted
should work, using GCC.

C++ has different language:

  7.2:

    For an enumeration where e_min is the smallest enumeration and
    e_max is the largest, the values of the enumeration are the values
    of the underlying type in the range b_min to b_max, where b_min
    and b_max are, respectively, the smallest and largest values of
    the smallest bit-field that can store e_min and e_max.

So, given:

  enum E { a = 2, b = 6 };

the values are actually { 0, 1, 2, 3, 4, 5, 6, 7 }, since the smallest
bit-field that holds 6 is actually 3 bits.  Thus, if the original
posters enumeration did not have a maximum value that was one less
than a power of two, the code would have been legal C++ as well.

But, I think the C9X language means that GCC is going to have to do what
the poster expects for C9X.  There's no reason to be incompatible
between the various C-like languages; all of them certainly permit the
looser C9X interpretation.  

The original patch looked reasonable to me, FWIW.

-- 
Mark Mitchell 			mark@markmitchell.com
Mark Mitchell Consulting	http://www.markmitchell.com