From:	Moogla <ngoldber AT lan DOT tjhsst DOT edu>
Newsgroups:	comp.os.msdos.djgpp
Subject:	Re: EUREKA... note about pcx header
Date:	Tue, 10 Aug 1999 23:13:41 -0400
Organization:	MindSpring Enterprises
Lines:	32
Message-ID:	<37B0EA65.3986@lan.tjhsst.edu>
References:	<Dd1s3.242$dr6 DOT 4824 AT news1 DOT rdc2 DOT on DOT home DOT com> <37B0BE23 DOT 2326F43C AT id-base DOT com> <37B0DFAD DOT 2AE5 AT surfsouth DOT com>
NNTP-Posting-Host:	a5.f7.46.7a
Mime-Version:	1.0
X-Server-Date:	11 Aug 1999 03:15:46 GMT
X-Mailer:	Mozilla 3.01 (Win95; I)
To:	djgpp AT delorie DOT com
DJ-Gateway:	from newsgroup comp.os.msdos.djgpp
Reply-To:	djgpp AT delorie DOT com

>   Most protected mode compilers compile based on a 32 bit integer.
> DJGPP does this.
>   For anyone who doesn't already know this (it is pretty common
> knowledge), in DJGPP:
>   char = 1 byte
>   short = 2 bytes
>   int = 4 bytes
>   long = 8 bytes
Sigh, sigh, sigh. No.

char = 1 byte always.
short = 2 byte always
long = 4 byte always
long long = 8 byte always (as of recently)
BUT *unspeficied variety* int in DJGPP is hardcoded to a long.


> 
>   And please don't ask about the time I took a new way to do fixed
> point math to a CS professor who thought it was brilliant and he sent
> me to a Computer Engineering prof who said, "Yes, congrats, you just
> rediscovered floating point."  That was a bad day.
> 

and for kicks:
short float = 32 bit IEEE float always (no matter what type of FPU)
double (float) = 64 bit IEEE float always
long double (float) = 80 bit IEEE float always
float (unspecified) in DJGPP has a size that can be defined by context
or compiler flags, but if it has a choice, 64-bit.

moogla

- Raw text -