From: Nate Eldredge <neldredge AT hmc DOT edu>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: Length of Chars...
Date: 22 Jan 2000 12:55:35 -0800
Organization: InterWorld Communications
Lines: 63
Message-ID: <83snzpq114.fsf@mercury.st.hmc.edu>
References: <3888ED7B DOT DF52FEB2 AT ou DOT edu> <38896068 DOT 8C5927C0 AT is DOT elta DOT co DOT il> <388A0530 DOT DF2B2F31 AT ou DOT edu>
NNTP-Posting-Host: mercury.st.hmc.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
X-Trace: nntp1.interworld.net 948574647 17320 134.173.45.219 (22 Jan 2000 20:57:27 GMT)
X-Complaints-To: usenet AT nntp1 DOT interworld DOT net
NNTP-Posting-Date: 22 Jan 2000 20:57:27 GMT
User-Agent: Gnus/5.0802 (Gnus v5.8.2) Emacs/20.5
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com

David Cleaver <davidis AT ou DOT edu> writes:

> OK, I'll explain a little 'bit' more.  :)  I'm writing a program to
> factor numbers, hopefully, very quickly.  Now, in order to do this I
> have created a bunch of two dimensional arrays whose elements only
> consist of zero's and one's.  Since this is the case, I decided to
> reduce the size of the array from using an int for each zero or one, and
> have packed all the zero's and one's into unsigned char's.  As you can
> see, this greatly reduced the amount of storage needed for these
> tables.  However, if the char type ever increases in size, my tables
> wouldn't be as tighly packed as possible, so I wanted to know if it was
> ever going to change.

I don't think you need to worry here.  AFAIK, `char' is defined to be
the smallest standard integer data type (not counting bitfields), so
even if `char' were increased in size, you wouldn't be able to do much
about it.

Also, judging from the amount of breakage that would result if a
compiler were to change the size, I suspect they'd include provision
for keeping the old behavior (a switch or something).

> Eli, what did you mean by using the "fastest data type" in a for loop? 
> What is the fastest data type that you are refering to here?  I wouldn't
> mind being able to speed up my program a bit more.

Most machines have one data type that is fastest to operate on,
normally the one that corresponds to the native wordsize.  On the 386,
it's the 32-bit int.  The 16-bit short, on the other hand, requires a
special instruction prefix to operate on, and each instruction
requires one more cycle than the corresponding 32-bit instruction.
Thus the code generated for `i++' will be both smaller and faster if
`i' is an int rather than a short.

So if `i' were a loop counter, one would want to make it `int' even if
the additional range wasn't needed.  The two bytes saved by using
`short' would be negligible, while the speed savings might not be.

Most machines define `int' as the data type which the machine handles
most efficiently, so it's usually your best bet.

> Also, is it ok to store unsigned chars in hex form?  I'll give an
> example below:
> 
> unsigned char array1[3][3] = {{0x2a, 0xf4, 0x08},
>                               {0x96, 0xa5, 0x7c},
>                               {0x8e, 0x59, 0x8b});
> 
> Is this ok, or will the compiler complain, or will I even still be able
> to access it like I normally would?  Again, thank you(everyone) for
> helping with all my questions.

Yes, it's perfectly legal.  Anywhere you want an integer value (even
if it's only the size of a char), you may specify it in decimal,
octal, hex, or character form.  So the following are all equivalent,
everywhere (AFAIK):

114, 0x72, 0162, 'r'

-- 

Nate Eldredge
neldredge AT hmc DOT edu