home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!gatech!purdue!mentor.cc.purdue.edu!pop.stat.purdue.edu!hrubin
- From: hrubin@pop.stat.purdue.edu (Herman Rubin)
- Newsgroups: comp.arch
- Subject: Lengths. Was: Re: 32 => 64 Transition
- Message-ID: <56918@mentor.cc.purdue.edu>
- Date: 15 Aug 92 14:34:48 GMT
- References: <1992Aug11.125326.16719@email.tuwien.ac.at> <id.UHAS.9TA@ferranti.com> <robert.713773090@cs.anu.edu.au>
- Sender: news@mentor.cc.purdue.edu
- Organization: Purdue University Statistics Department
- Lines: 41
-
- In article <robert.713773090@cs.anu.edu.au> robert@cs.anu.edu.au (Robert Cohen) writes:
- >peter@ferranti.com (peter da silva) writes:
-
- >>What size would *you* make an int? 64 bits? What do you do for a 32-bit
- >>data type? Short? Then what do you do for a 16-bit data type? Create another
- >>abomination "short short" or "long char"? "long long" is bad enough.
-
- >The perfect solution: int= 64, long =128 (for real men's numbers),
- >short = 32, char = 8, long char= 16, short short = 24,
- >long short = 48 and short long = 96.
- >If we need more precision we can start using short short long and
- >long short long etc.
- >In case anyone hasn't realised yet :-) :-) :-)
-
- >More seriously, I belive that we need a way to machine independently
- >specify the precision such as int:32.
- >When I write a program I usually know what sort of range I expect
- >a variable to have to hold. I dont care if its an int or a long or a short
- >on different machines, I want a number about that big.
-
- >Who uses 16 bit variables anymore anyway, memory's cheap, right :-)
-
- There are many situations in which one generates what is essentially
- a bit stream, and works with that. The generation process takes about
- the same length of time regardless of the sizes of the "words" produced
- up to machine arithmetic size. I have not found any situations yet
- where using 16-bit blocks is the "right" thing to do, but some situations
- which I have not completely investigated may call for it.
-
- There is another usage which will be coming out shortly, that of text
- communication not restricted to the Latin alphabet. I have read on other
- groups about a protocol, called unicode, which will use 16 bits/character.
- As you have said, memory's cheap, so why should we be limited to 8 bits?
-
- Other sizes which should be considered are 52 (53?), the length of the
- significand in an IEEE float, etc.
- --
- Herman Rubin, Dept. of Statistics, Purdue Univ., West Lafayette IN47907-1399
- Phone: (317)494-6054
- hrubin@pop.stat.purdue.edu (Internet, bitnet)
- {purdue,pur-ee}!pop.stat!hrubin(UUCP)
-