Likelihood That (CHAR_BIT == 8). - c++

On 10 ???, 07:23, Le Chaud Lapin <jaibudu...@gmail.com> wrote:
> Hi All,
>
> I would like to know of existing systems where CHAR_BIT of <climits>
> is not equal to 8. I am particular interested in prevalent systems.
>
> Technically, my code is mostly portable, but only within a particular
> machine. Innter-machine interaction (serialization for example), there
> will be problems for those machines where CHAR_BIT != 8.

Likelihood depends on what you are targeting. But, to give a few
specific examples - some DSPs cannot operate on individual octets, so
char is whatever the word size of the architecture is - e.g. 32-bit.
There was a link to documentation for a C++ compiler for such a
platform (SHARC) given earlier in a thread titled "mixed-sign
arithmetic and auto" (which I would recommend to read, by the way - it
has some more information somewhat relevant to the topic) - here it
is:

http://www.****og.com/UploadedFiles/...ARC_cc_man.pdf

Apparently, this one has 32-bit char, and, as a consequence,
sizeof(int)==sizeof(char)==1


--
[ See http://www.gotw.ca/resources/clcm.htm for info about ]
[ comp.lang.c++.moderated. First time posters: Do this! ]

On 10 Maj, 05:23, Le Chaud Lapin <jaibudu...@gmail.com> wrote:
> Hi All,
>
> I would like to know of existing systems where CHAR_BIT of <climits>
> is not equal to 8. I am particular interested in prevalent systems.
>
> Technically, my code is mostly portable, but only within a particular
> machine. Innter-machine interaction (serialization for example), there
> will be problems for those machines where CHAR_BIT != 8.
>
> Yes, yes, I know, it enough effort, I can rewrite my code to eliminate
> any potential compatibilities even then, but still, it helps to
> know.

Probably a lot of CPU's have CHAR_BIT = 16 or 32. If you plan for
porting your code to e.g. cellphones and PDAs you are likely to get
into troubles (I've programmed on a PDA that had CHAR_BIT = 16).
If you turn into more exotic architectures you are of course even more
likely to experience problems.

/Peter

--
[ See http://www.gotw.ca/resources/clcm.htm for info about ]
[ comp.lang.c++.moderated. First time posters: Do this! ]

Le Chaud Lapin wrote:
> Hi All,
>
> I would like to know of existing systems where CHAR_BIT of <climits>
> is not equal to 8. I am particular interested in prevalent systems.
>
> Technically, my code is mostly portable, but only within a particular
> machine. Innter-machine interaction (serialization for example), there
> will be problems for those machines where CHAR_BIT != 8.
>
> Yes, yes, I know, it enough effort, I can rewrite my code to eliminate
> any potential compatibilities even then, but still, it helps to
> know.

There are still a few 36-bit DEC machines running
(try Telnet to "DEC-10.PDPplanet.COM") and some Unisys
B series mainframes running 36-bit OS/2200. But when Unisys
discontinued the ClearPath 36-bit server line last year, in favor of software
emulation of the 36-bit machines for old code, commercial production of the
last non-byte-oriented hardware ended.

So you probably don't have to support char sizes other than 8.

Here, for fun, is a useful piece of C code written to work portably across
7, 8, and 9-bit character sizes. This tests passwords for "obviousness"
without needing a dictionary file.

http://www.animats.com/source/obvious/obvious.c

John Nagle



--
[ See http://www.gotw.ca/resources/clcm.htm for info about ]
[ comp.lang.c++.moderated. First time posters: Do this! ]

Le Chaud Lapin wrote:
> Hi All,
>
> I would like to know of existing systems where CHAR_BIT of <climits>
> is not equal to 8. I am particular interested in prevalent systems.
>
> Technically, my code is mostly portable, but only within a particular
> machine. Innter-machine interaction (serialization for example), there
> will be problems for those machines where CHAR_BIT != 8.

Some TI DSPs have CHAR_BIT==16.

So long,
Thomas

--
[ See http://www.gotw.ca/resources/clcm.htm for info about ]
[ comp.lang.c++.moderated. First time posters: Do this! ]

Le Chaud Lapin wrote:

> Hi All,
>
> I would like to know of existing systems where CHAR_BIT of <climits>
> is not equal to 8. I am particular interested in prevalent systems.

I am currently developing software for the BlueCore-5 chip (a Bluetooth
chip from Cambridge Silicon Radio, www.csr.com) which has CHAR_BIT ==
16.

>
> Technically, my code is mostly portable, but only within a particular
> machine. Innter-machine interaction (serialization for example), there
> will be problems for those machines where CHAR_BIT != 8.

The communication might not be a problem for the BlueCore chip. It often
uses octet-oriented communication mechanisms (UART and I2C) and handles
the conversion between byte and octet internally.
Other processors that often interface over octet-oriented mechanisms
might do the same.

>
> Yes, yes, I know, it enough effort, I can rewrite my code to eliminate
> any potential compatibilities even then, but still, it helps to
> know.
>
> TIA,
>
> -Le Chaud Lapin-
>
Bart v Ingen Schenau
--
a.c.l.l.c-c++ FAQ: http://www.comeaucomputing.com/learn/faq
c.l.c FAQ: http://c-faq.com/
c.l.c++ FAQ: http://www.parashift.com/c++-faq-lite/

[ See http://www.gotw.ca/resources/clcm.htm for info about ]
[ comp.lang.c++.moderated. First time posters: Do this! ]

On May 10, 8:14 am, John Nagle <na...@animats.com> wrote:
<snip>

> There are still a few 36-bit DEC machines running
> (try Telnet to "DEC-10.PDPplanet.COM") and some Unisys
> B series mainframes running 36-bit OS/2200. But when Unisys
> discontinued the ClearPath 36-bit server line last year, in favor of
software
> emulation of the 36-bit machines for old code, commercial production of
the
> last non-byte-oriented hardware ended.

Has production of supercomputers ended, too?

Supercomputers generally don't have (hardware) access
for units smaller than one word, for efficiency reasons.
E.g., CDC 6600/7600 series, Crays, etc.

I used to run C++ programs on supercomputers, and there
is a subgroup of the C++ standards group interested in
C++ for numerical applications, so they are
relevant to this group.

> So you probably don't have to support char sizes other than 8.

I believe Cray (64-bit word) compilers emulate(d) 8-bit
access, so "char" is still 8 bits. It does mean that
handling of "char" variables and arrays is extremely
inefficient, but byte-oriented computation forms an
infinitesimal part of these machines' workload, so it
doesn't matter.

The CDC machines have/had 60-bit words. I never did
any C programming on them (let alone C++), so I don't
know if "char" was 6, 8, or 10 bits on them. The
CDC operating systems used 6-bit characters, but also
sometimes stored 15 8-bit characters in 2 words.

I don't know if they are still manufacturing this
architecture, though.

--
[ See http://www.gotw.ca/resources/clcm.htm for info about ]
[ comp.lang.c++.moderated. First time posters: Do this! ]