Received: from sloth.swcp.com (sloth.swcp.com [198.59.115.25]) by nacm.com (8.6.9/8.6.9) with ESMTP id WAA25831 for <executor@nacm.com>; Mon, 7 Nov 1994 22:18:02 -0800
Received: from iclone.UUCP (uucp@localhost) by sloth.swcp.com (8.6.9/8.6.9) with UUCP id XAA14341 for nacm.com!executor; Mon, 7 Nov 1994 23:19:20 -0700
Received: by iclone (NX5.67d/NX3.0M)
id AA23684; Mon, 7 Nov 94 21:49:12 -0700
Date: Mon, 7 Nov 94 21:49:12 -0700
From: "Mathew J. Hostetter" <iclone!mat@sloth.swcp.com>
Message-Id: <9411080449.AA23684@iclone>
To: executor@nacm.com
Subject: Re: Another Question
Sender: Executor-Owner@nacm.com
Precedence: bulk
Jeff Tupper <mooncake@csri.toronto.edu> writes:
>What does executor do for SANE? Does it emulate all of the SANE calls?
>What I am interested in is whether it supports control of the rounding
>direction. A program I sell for the Mac uses this feature so it is not
>as obscure as it may at first seem... (In some sense I am an outsider
>here since I don't have an IBM yet and haven't tried executor... but I
>will be getting an IBM in a couple of months and will try it then :)
Our SANE implementation uses the 80387 hardware (or an 80387 emulator
if an FPU isn't present) to do 80-bit floating point arithmetic. We
implement every SANE call, although some of them are only very close
approximations to the official specs and not bit-for-bit accurate.
Fortunately, very few programs care about strict SANE conformance.
Those that do probably won't work correctly.
To answer your question, when the emulated program changes the SANE
rounding mode, Executor changes the 80387 rounding mode in the same
way. As far as we know, your program should work, but to be honest
we've got so many things to work on that we haven't spent much time
testing floating point details. We'd like to hear about any problems
