A. Paul R. Cooper
British Antarctic Survey

*****          Any opinions given here are my own,          *****
***** and not necessarily those of British Antarctic Survey *****

Kurt
--
| Kurt Watzka                             Phone : +49-89-2180-6254

Because "the sizeof operator yields the size (in bytes) of its operand" and
the result
is1 for the char types, a char object always occupies exactly one byte of
storage,
not "2 bytes, or 4 bytes, or 256 bytes," at least using Standard C's
definition of a
"byte."

When talking about bytes in the context of C, you have to be very careful. Do
you
mean simply 8-bits or the formal, more elusive definition provided in the
Standard?

Hans Friedrich Steffani
--
Hans Friedrich Steffani
Institut fuer Elektrische Maschinen und Antriebe
TU Chemnitz-Zwickau

HOWEVER!

I would still do bytewise I/O using fgetc/fputc, because some characters are
special, not just to C, but to the operating system. In particular,
end-of-line is amazingly system specific. UNIX filesystems follow one
convention, DOS another, various IBM operating systems follow several
conventions all at once depending on file header settings! I would say that my
distinction between characters and bytes is valid, because a character need
not map one-to-one onto a byte. A character may be none, one or many bytes. A
byte is a byte (unless you remember some old systems with 7 bit bytes....)

At a slightly higher level, I am using DEC Alpha workstations at the moment. I
keep coming accross major problems with commercial software where people have
obviously made assumptions about the wordlength of the host machine. It is
amazing how many people have written software which ASSUMES that an int is
two bytes!

A. Paul R. Cooper
British Antarctic Survey

*****          Any opinions given here are my own,          *****
***** and not necessarily those of British Antarctic Survey *****

8.10:   I'm starting to think about multinational character sets, and
        I'm worried about the implications of making sizeof(char) be 2
        so that 16-bit character sets can be represented.

A:      If type char were made 16 bits, sizeof(char) would still be 1,
        and CHAR_BIT in <limits.h> would be 16, and it would simply be
        impossible to declare (or allocate with malloc()) a single 8-bit
        object.

        Traditionally, a byte is not necessarily 8 bits, but merely a
        smallish region of memory, usually suitable for storing one
        character.  The C Standard follows this usage, so the bytes used
        by malloc() and sizeof can be more than 8 bits. [1]  (The
        Standard does not allow them to be less.)

        To allow manipulation of multinational character sets without
        requiring an expansion of type char, ANSI/ISO C defines the
        "wide" character type wchar_t, and corresponding wide string
        literals, and functions for manipulating and converting strings
        of wide characters.

        References:
        ANSI Sec. 2.2.1.2, Sec. 3.1.3.4, Sec. 3.1.4,
                Sec. 4.1.5, Sec. 4.10.7, Sec. 4.10.8
        ISO Sec. 5.2.1.2, Sec. 6.1.3.4, Sec. 6.1.4, Sec. 7.1.6,
                Sec. 7.10.7, Sec. 7.10.8
        Rationale Sec. 2.2.1.2
        H&S Sec. 2.7.3 pp. 29-30, Sec. 2.7.4 p. 33, Sec. 11.1 p. 293,
                Secs. 11.7,11.8 pp. 303-310
__________
1.  Formally, bytes of exactly 8 bits are referred to as "octets."

                                                Steve Summit

The C definition also says that a char has >= 8 bits, NOT == 8 bits.
The same would also apply to a byte.

                                        Gordon L. Burditt
                                        sneaky.lonestar.org!gordon

: By the C definition, sizeof(char) == 1, and the units of sizeof are
: in bytes.  Therefore a byte and a char have the same size.
: A char is a recognized C type; a byte is not.

I am not an expert in the field of compiler nor programming.
But I want to ask whether "byte" can have "signed" and "unsigned" properties,
like those of "char", that contribute to many portablilty problems and bugs.
IMO, "byte" is always signed. Right? or is it machine/implementation dependent?

P.S. I have never use the type "byte" in my C programs, pls. forgive me
if my question is too silly :>

--
Peter Wu
E8-P)

--
miracles will have their claimers
more will bow to Rome
he and she are in the house
but there's only me at home

--
  Lawrence Crowl                503-737-2554    Computer Science Department

              http://www.cs.orst.edu/~crowl/    Corvallis, Oregon,  97331-3202