Path: utzoo!attcan!uunet!mailrus!cs.utexas.edu!usc!brutus.cs.uiuc.edu!uakari.primate.wisc.edu!ames!amdahl!pacbell!att!mcdchg!ddsw1!nvk
From: nvk@ddsw1.MCS.COM (Norman Kohn)
Newsgroups: comp.unix.microport
Subject: Re: fp emulation in microport DOS MERGE
Message-ID: <1990Jan11.060750.16038@ddsw1.MCS.COM>
Date: 11 Jan 90 06:07:50 GMT
References: <1990Jan6.230548.23780@ddsw1.MCS.COM> <10@cimcor.MN.ORG>
Reply-To: nvk@ddsw1.MCS.COM (Norman Kohn)
Organization: ddsw1.MCS.COM Contributor, Mundelein, IL
Lines: 29

In article <10@cimcor.MN.ORG> mike@cimcor.MN.ORG (Michael Grenier) writes:
>From article <1990Jan6.230548.23780@ddsw1.MCS.COM>, by nvk@ddsw1.MCS.COM (Norman Kohn):
>% Microport's dos merge kernel seems to have some problem with
>% 80387 emulation.  On a machine with no math coprocessor the code
>% #define MATHTYPE int
>% main()
>% {
>% 	MATHTYPE x;int j;
>% 	j=2;
>% 	x=(MATHTYPE) j;
>% 	j=(int) x;
>% 	printf("%d\n",j);
>% }
>% fails (produces results of 0) if MATHTYPE is float or double.
>% sdb confirms that the conversions are failing.
>% This may suggest why awk, which uses floating point heavily, gets very confused
>% under DOSMERGE.
>
>Strange, the above code worked fine on this machine. The previously
>posted awk script also ran fine on this 386 DosMerge Kernel
>without an FP. I strongly suggest a CPU board problem is
>at fault here.
>
It would seem so, yet everything works fine with the 3.0e kernel.
I have the first post-beta MERGE release (I don't know its
number) and am awaiting uport's more recent update.  Perhaps
you're running a more recent version?

-- 
Norman Kohn   		| ...ddsw1!nvk
Chicago, Il.		| days/ans svc: (312) 650-6840
			| eves: (312) 373-0564