Yesterday someone ask me: "How was it determined that 1 byte would actually
(usually) be equal to 8 bits?  Why not 9 bits or 7 bits?"  I didn't know the
answer.  Do you?

Yup.

Because 8 is the smallest integral power of 2 that can hold all the
characters of C's character set.

(Okay, it's a wild guess.  Does anyone have a better idea?)

Gregory Pietsch

It isn't 8 bits, as far as the C language is concerned.  The number of bits in
a char/byte in the C language is CHAR_BIT bits, as defined in limits.h.
There are definitely systems on which it is 9 or more bits (up to at least 32
on a popular DSP).

As far as C is concerned, it cannot be 7 bits, since that would not allow the
implementation of conforming characters.  Hence, it is _at least_ 8 bits, but
possibly more.
--
C-FAQ: http://www.eskimo.com/~scs/C-faq/top.html
 "The C-FAQ Book" ISBN 0-201-84519-9
C.A.P. FAQ: ftp://cap.connx.com/pub/Chess%20Analysis%20Project%20FAQ.htm

Charles Martin a crit dans le message ...

How is this a C question ?

<slightly ot>
It depences the target. A byte (aka char in C) is the smallest amount of
adressable memory. From the C language point of view, a byte must have at
last 8 bits.

Some targets have bytes of 8,9,16,32,36 bits, whatever...

Note: An 8-bit data is called an octet.
</ot>

--
-hs-    Tabs out, spaces in.
CLC-FAQ: http://www.eskimo.com/~scs/C-faq/top.html
ISO-C Library: http://www.dinkum.com/htm_cl
FAQ de FCLC : http://www.isty-info.uvsq.fr/~rumeau/fclc

At the beginning, the memory systems were not byte-addressed (each address
was a word address) and the words were divided in bytes.  The sizes of
these bytes were typically 6, 7 or 9 bits (the typical words were 12, 18
and 36 bits).  Then, IBM came with the byte addressed 360 that had
8-bit bytes.  A few years later, DEC came with the byte addressed
PDP-11 that had 8-bit bytes.  Some three or four years after that Intel
came with the 8080, an 8-bit microprocessor.  Each of these systems was
the most popular in its class (mainframe, mini and micro) at the time,
so the rest of the industry followed the trend.

By the time C was standardised, 8-bit bytes were the de facto industry
standard, so the ANSI committee that did the job decided to require
that the C byte has *at least* 8 bits, although the minimum number of
bits required by the C character set is 7.

Dan
--
Dan Pop
CERN, IT Division

Mail:  CERN - IT, Bat. 31 1-014, CH-1211 Geneve 23, Switzerland

It wasn't determined. A byte could be 9 bits or 7 bits.  In the C language, a
byte is at least 8 bits wide; the standard specifies that a byte is the
smallest individually addressable unit of storage, and that it is at least
eight bits wide.

You are confusing octets and bytes. An octet is a unit of information
consisting of 8 ordered bits. It is a common unit of data transfer in
networking. Since C bytes are 8 *or more* bits wide, the type unsigned
char can always store the binary value of an octet.

--
Any hyperlinks appearing in this article were inserted by the unscrupulous
operators of a Usenet-to-web gateway, without obtaining the proper permission
of the author, who does not endorse any of the linked-to products or services.

Yes. The two nibbles of an 8 bit byte can each store a BCD digit.
Also, an 8 bit byte can represent an ASCII character plus a parity bit.  

This is good topic for alt.folklore.computers; I would guess that if
that newsgroup has a FAQ, this would probably be in it.

--
Any hyperlinks appearing in this article were inserted by the unscrupulous
operators of a Usenet-to-web gateway, without obtaining the proper permission
of the author, who does not endorse any of the linked-to products or services.

On Wed, 1 Nov 2000 16:42:47 -0800, "Dann Corbit"

Actually on at least two DSPs that I know of, TI and AD.

Jack Klein
--
Home: http://jackklein.home.att.net

I am intrigued.

My fallible understanding of history was that, in the beginning, there
was the bit (binary digit) and there was the word.

The word size was architecture dependent, and common values included
4, 8, 12, 16, 24 and 32 bits.

However, the 8-bit unit prooved so useful (ASCII = 7 bits + parity,
two hex digits, etc) that it became a widespread common denominator,
irrespective of machine word size. So widespread that it needed a
name.

I didn't think the term 'byte' existed in the C standard at all - we
simply have a coincidental correlation between the most common
representations of a character and what we independently think of as a
byte. Character sizes can obviously be bigger than 8 bits (unicode,
for instance) but why redefine the byte - the word was meant to be
redefined, but the byte is a kind of concensus-based standard that is
too useful to arbitrarily break.

--
Steve Horne

<snip "how many bits per byte?">

I hope you will allow me to show you that the term 'byte' not only
*exists* in the C Standard, but also is in fact *defined* (for the
purposes of C programmers) in that Standard. The definition that follows
is from the n869 draft of ISO/EIC 9899:1999, and as far as I know it's
identical to the wording in the current Standard (but I can check at the
weekend if need be):

       3.4
       [#1] byte
       addressable  unit  of  data storage large enough to hold any
       member  of  the  basic  character  set  of   the   execution
       environment

And, a little later, we have:

       5.2.4.2  Numerical limits

       [#1] A conforming  implementation  shall  document  all  the
       limits  specified  in this subclause, which are specified in
       the headers <limits.h> and <float.h>.  Additional limits are
       specified in <stdint.h>.

       5.2.4.2.1  Sizes of integer types <limits.h>

       [#1]  The  values  given below shall be replaced by constant
       expressions  suitable   for   use   in   #if   preprocessing
       directives.   Moreover,  except for CHAR_BIT and MB_LEN_MAX,
       the following shall be replaced by expressions that have the
       same  type  as  would an expression that is an object of the
       corresponding  type  converted  according  to  the   integer
       promotions.   Their  implementation-defined  values shall be
       equal or greater in  magnitude  (absolute  value)  to  those
       shown, with the same sign.

         -- number  of  bits for smallest object that is not a bit-
            field (byte)
            CHAR_BIT                         8

Thus, the C Standard defines a byte to be at least 8 bits in size (note
the wording - the value of CHAR_BIT is implementation-defined, and must
have a magnitude of at least 8 (and a positive sign!)).

<shrug> If you want to know /why/ the C Standard "redefines" the byte,
you may care to check with the guys in comp.std.c, since several ISO C
committee guys subscribe to that group. They rarely poke their noses
into comp.lang.c, because they think we're too loud. :-)

--
Richard Heathfield
"Usenet is a strange place." - Dennis M Ritchie, 29 July 1999.
C FAQ: http://www.eskimo.com/~scs/C-faq/top.html
K&R Answers: http://users.powernet.co.uk/eton/kandr2/index.html

On Thu, 02 Nov 2000 13:11:29 +0000, Richard Heathfield

OK - I wasn't aware of that.

Here I was wrong, though - see the 7 and 9 bit bytes that others have
mentioned. The standards guys presumably just set a constraint on
something that might otherwise have been variable.

Oh - thats the English meaning of the word variable, not a variable as
defined in the C standard  ;-)

--
Steve Horne

<snip>

Er... the C Standard doesn't define the term "variable". :-)

--
Richard Heathfield
"Usenet is a strange place." - Dennis M Ritchie, 29 July 1999.
C FAQ: http://www.eskimo.com/~scs/C-faq/top.html
K&R Answers: http://users.powernet.co.uk/eton/kandr2/index.html

Not in the text, no -- but it does use the term, and thus presumably
falls back on the definition in one of its references, probably the
first of these two:

        ISO/IEC   2382-1:1993,   Information   technology    -
        Vocabulary - Part 1: Fundamental terms.

        ISO/IEC TR 10176, Information technology -  Guidelines
        for the preparation of programming language standards.

(I do not have either one so I cannot check.)
--
In-Real-Life: Chris Torek, Berkeley Software Design Inc

And 7, 9 and other values. powers-of-two only became popular when
people _stopped_ doing parity properly (gd&r).

you do realise that your PC memory is probably NINE bit don't you?  8
plus parity...

Byte existed before the "standardization" onto 8 bits. Don't let
mainframe people hear you call 8 bits be a standard by the way.

As Richard pointed out... it is.

--
Mark McIntyre
C- FAQ: http://www.eskimo.com/~scs/C-faq/top.html