Summary of changes between version 1.0beta1 and version 0.18beta1
==========================================================
<Christoph Lauter>
        Large performance improvement in asin and acos, now in asincos.c
        Added asinpi and acospi
Summary of changes between version 0.18beta1 and version 0.17beta1
==========================================================
<Florent de Dinechin> 
	Added atanpi

Summary of changes between version 0.17beta1 and version 0.14beta1
==========================================================
<Florent de Dinechin> 
	Added sinpi, cospi, tanpi
<Christoph Lauter>
        Added power, currently unproven
	
Summary of changes between version 0.14beta1 and version 0.13beta1
==========================================================
<Christoph Lauter>
	Added log1p
	Added tests/computemd5list.sh
<Florent de Dinechin> 
	Documentation updates, added more worst cases
<Alex Bernier and Florent de Dinechin>
	Interval functions for exp and log (in the CVS, not yet in the release)

Summary of changes between version 0.13beta1 (never released) and version 0.11beta1
==========================================================
<Christoph Lauter>
	Added expm1, acos
	Corrected small possible bug in asin 

Summary of changes between version 0.11beta1 and version 0.10beta
==========================================================
<Christoph Lauter>
	Added asin
	More Gappa
<Florent de Dinechin>
        Correction of a serious bug (found by Paul Zimmermann) 
  	  in the computation of rounding constants. 
	  (Fortunately the bug was in the proof, the code was OK)
	More Gappa proofs
	
Summary of changes between version 0.10beta and version 0.8
==========================================================
<Christoph Lauter>
	Added log2 and log10 using triple double (log2-td.*, log10-td.*)
	Triple-double versions of exp and log (exp-td.*, log-td.*)
<Florent de Dinechin>
	log-de, using double-extended, much faster on IA32 and IA64
	Cleaning up, autotoolization of the new functions
	Proofs using the Gappa tool
	
Summary of changes between version 0.8 and version 0.8beta4
==========================================================
<David Defour>
	Cleaning up and fixes to improve portability
<Florent de Dinechin>
	Finished proofs of trigonometric functions
	Some cleaning up and documentation improvement
	Fixed bug for large arguments to sinh and cosh
	Added crlibm.spec (to build RPMs)
	
Summary of changes between version 0.8beta4 and version 0.8beta3
==========================================================
       
<Florent de Dinechin>
        Added crlibm_generate_test_vectors
	Added test vectors to all functions, including some 
	  of Lefevre/Muller worst cases
	Cleant up the accurate trig functions, now in trigo.c
	Cleant up cosh and sinh
	Added Gappa input file to assist the proof of the trig functions
<Patrice Pelissier>
	Added AMD64 architecture to autoconf stuff
	Improved timings on x86


Summary of changes between version 0.8beta3 and version 0.8beta2
==========================================================
       
<Florent de Dinechin>
        Added crlibm_blindtest and automatic testsuite
	Consequently corrected quite a few bugs in extreme cases
	Cleant up cosh and sinh

Summary of changes between version 0.8beta2 and version 0.8beta
==========================================================
       
<David Defour and Florent de Dinechin>
	Rewritten faster trigonometric functions
<Florent de Dinechin>
        Added crlibm_blindtest and automatic testsuite 
        Added crlibm_exit()
	crlibm_init() now returns the old processor status flag
	Proof for the trigonometric functions in progress
<Philippe Defert>
	Added crlibm.spec file to build rpms  to the CVS
	
Summary of changes between version 0.8beta and version 0.2:
==========================================================

<Catherine Daramy & Florent de Dinechin>
	Added functions : sin cos tan

<Nicolas Gast>
	Added function atan 
	Added function atan specialized for pentium (atan-pentium.c)

<Matthieu Gallet>
	Added functions sinh cosh

<Florent de Dinechin>
	Changed Maple files from .mw and .mws to .mpl
	Added FMA support in crlibm_private.h
	Added polish to some of the proofs
	Improved performance tests for tighter timing, 
	Fairer speed comparison to mpfr: We now compare to mpfr with
	 53-bit mantissa, although this doesnot provide correct rounding
	 if the result is a denormal (double rounding). Previously we
	 compared to 150-bit mpfr.

Version 0.2 : exp and log

