Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!wuarchive!sdd.hp.com!think.com!snorkelwacker.mit.edu!shelby!unix!hplabs!hpfcso!cunniff
From: cunniff@hpfcso.HP.COM (Ross Cunniff)
Newsgroups: comp.sys.amiga.programmer
Subject: Re: SAS Bugs
Message-ID: <101060007@hpfcso.HP.COM>
Date: 4 Feb 91 18:35:38 GMT
References: <91028.004246T32QC@CUNYVM.BITNET>
Organization: Hewlett-Packard, Fort Collins, CO, USA
Lines: 33

>>Well, this weekend, I just got bitten by the following nasty bug ...

>>	func( a, b )
>>	float a;
>>	long *b;
>>	{
>>	    *(float *)b = a;
>>	}
>>
>> [ explains how compiler thinks a is the size of a double ]

> This is NOT a bug.  This is how C used to work.  It bit me once
> several years ago.  Before the ANSI changes to C the only way to pass
> floating point numbers to a routine was as a double (never as a
> float).  Since you are using the old syntax then when you declare "a"
> as a float the compiler will "assume" it is a double since C used to
> only pass things as doubles.  The old syntax is kept for backward
> compatibility.

It most certainly IS a bug; you didn't read the assembly code.
The compiler thought the location of a was 4 bytes past the location
of b, not 8 as it should have been.  The code is old, so there was
no prototype present anywhere.  And yes, I already know how to work
around it.

By the way, I have already received a reply from a representative
of SAS who has provided me with some great support.  I have no
complaints with the company; all compilers have bugs, even the
ones I write...

				Ross Cunniff
				Hewlett-Packard Colorado Language Lab
				cunniff@hpfcla.HP.COM