Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!seismo!rutgers!ames!elroy!smeagol!jplgodo!wlbr!scgvaxd!stb!michael
From: michael@stb.UUCP (Michael)
Newsgroups: news.admin,news.groups
Subject: Re: Easy way to remove binary groups
Message-ID: <1693@stb.UUCP>
Date: Wed, 29-Jul-87 13:37:18 EDT
Article-I.D.: stb.1693
Posted: Wed Jul 29 13:37:18 1987
Date-Received: Sat, 1-Aug-87 05:09:15 EDT
References: <266@brandx.rutgers.edu> <8225@utzoo.UUCP> <272@brandx.rutgers.edu> <2524@hoptoad.uucp> <174@genghis.UUCP>
Reply-To: michael@stb.UUCP (Michael)
Organization: STB BBS, La, Ca, USA, 90402
Lines: 34
Xref: mnetor news.admin:758 news.groups:1301

In article <174@genghis.UUCP> sns@genghis.UUCP (Sam Southard) writes:
>As long as I'm here, I might as well put my two cents in.  I do most, if not
>all, of my work on a PC/AT.  A lot of the sources that are put out make some
>assumptions about sizeof(int) among other things.  They won't work on the AT
>with the Lattice C or Microsoft C under DOS or the Xenix C compiler (I have
>all three).  It's not the fault of the compiler, but the assumptions of the
>author (to avoid flames, I'd like to be able to make such assumptions, and
>they are not unreasonable.  Unfortunately, the are also not valid for the AT).

Here's an excellent solution to this problem (and one I wish the IBM compilers
had as an option, at least): Make sizeof int == sizeof long == sizeof char*
== sizeof void*. 
 
 In other words, if you need 32 bits for a long, then ints will also be 32
bits. Pointers need (if I understand correctly how the segmemting works)
48 bits (32 offset, 16 for seqment reg). So, make all ints/longs reserve
48 bits of data and only use 32.
 
 PRESTO! You can assign ints/longs/char *'s interchangibly. You can forget
to force that NULL to a (char *)NULL and it will still work.

Realize: C is NOT ASSEMBLY (except on the pdp-11 :-). It is a language.
Whether that implementation is super-optimized or is portable is a
decision made by the compiler author. Commercial ones go for the efficiency.
PD ones can go for portability.

(in case you're curious: I have a C compiler for a 68000 that has a flag
for 16 bit, efficient ints vs. 32 bit, compatible ints. Since the O/S
was written with a different compiler (32 bit ints), the option was the
only way to go).

-- 
: Michael Gersten		seismo!scgvaxd!stb!michael
: Copy protection? Just say Pirate! (if its worth pirating)