Date: Mon, 28 Sep 1998 12:46:22 -0500
From: Eric Rudd <rudd AT cyberoptics DOT com>
Subject: libc math function upgrade work
To: djgpp-workers AT delorie DOT com
Cc: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
Reply-to: rudd AT cyberoptics DOT com
Message-id: <360FCB6E.44B7D417@cyberoptics.com>
Organization: CyberOptics
MIME-version: 1.0
Content-type: text/plain; charset=us-ascii
Content-transfer-encoding: 7bit
References: <Pine DOT SUN DOT 3 DOT 91 DOT 980928184446 DOT 1243C-100000 AT is>
Precedence: bulk

I have been working on replacements for the math functions in libc.  The
objectives were to bring the functions into strict ANSI compliance and
improve the accuracy of the routines, while maintaining the speed of the
existing libc routines.  The new routines are not quite finished yet, but I
have coded all of them except pow(), and in accuracy tests against some
384-bit reference routines that I downloaded from the Web, they have so far
been flawless (apart from range reduction being good only to 64 bits).

I have been in correspondence with Eli Zaretskii about this effort, and he
suggested that I post this message.  I would appreciate any suggestions that
you may have about putting the package together and submitting it.  Eli also
mentioned that there was a new test suite in the 2.02 alpha release, so will
download it and test my routines against that suite, in addition to the other
tests I have already run.

-Eric Rudd
rudd AT cyberoptics DOT com