1Version 2.x.x
2	* Bumped version to track OpenEXR
3	  (Piotr Stanczyk)
4
5Version 2.0.1
6	* Bumped version to track OpenEXR
7	  (Piotr Stanczyk)
8
9Version 2.0.0
10	* Bumped version to track OpenEXR
11	  (Piotr Stanczyk)
12	* Numerous minor fixes, missing includes etc
13
14Version 1.1.0.beta.1
15	* Added new module PyIlmBase : python bindings for IlmBase
16	  (Nick Rasmussen)
17	* Added git specific files
18	  (Piotr Stanczyk)
19	* Minor fixes for newer gcc versions and OS X.
20	  (misc)
21	* Preparation for OpenEXR v2 release { remove message for final release }
22	  (Piotr Stanczyk)
23        * Updated the so verison to 10
24	  (Piotr Stanczyk)
25	* Initial use of the CMake build system
26	  (Nicholas Yue)
27
28Version 1.0.3
29        * Added support for enabling/disabling large stack optimisations, used in
30          halfFunction.h.
31          (Piotr Stanczyk)
32	* Added ImathNoise.(h/cpp) files. Initializes Perlin noise to match the
33	  Renderman implmenetation.
34	  (Pixar Contribution)
35	* Fixed a number of missing includes to comply with stricter
36	  enforcement by gnu compilers.
37	  (Piotr Stanczyk)
38	* Depracated compiler flag: -Wno-long-double since it is no longer
39	  supported under OS X.
40	  (Piotr Stanczyk)
41	* A minor API change to Imath::Frustum has been made:  the functions
42	  'near' and 'far' have been renamed to 'nearPlane' and 'farPlane' due
43	  to conflicts with certain windows headers.  The former alternate
44	  accessor names for these values on windows ('hither' and 'yon')
45	  remain, though should be considered deprecated.
46	  (David Lenihan)
47	* Added SVD, eigenvalue solver, and procrustes fit calculations
48	  to ImathMatrixAlgo.
49	  (Chris Twigg, Ji Hun Yu)
50	* Added Imath::FrustumTest for frustum visibility testing.
51	  (Eric Johnston)
52	* Fixed a stack corruption in the matrix minorOf functions.
53	  (Nick Rasmussen)
54	* Visual studio 2008 project files have been added to the vc/vc9
55	  directory, and several minor visual studio compile fixes have
56	  been applied.
57	  (Nick Rasmussen)
58	* Updated the so verison to 7.
59	  (Piotr Stanczyk)
60	* Depracated the MacCode_Warrior and Shake submodules.
61	  (Piotr Stanczyk)
62
63Version 1.0.2
64	* Added support for targetting builds on 64bit Windows and minimising
65	  number of compiler warnings on Windows. Thanks to Ger Hobbelt for his
66	  contributions to CreateDLL.
67	  (Ji Hun Yu)
68	* Removed a spurious restrict qualifier in the matrix manipulation code
69	  that was causing the 64-bit MS compiler to generate code in release
70	  mode that caused incorrect results.
71	  (Ji Hun Yu)
72	* Added patches for improving universal binaries on OS X. Thanks to
73	  Paul Schneider for the contribution
74	  (Piotr Stanczyk)
75	* Imath::Box optimization: remove loops from methods by partially
76	  specializing the class, for boxes of two and three dimensions.
77	  (Piotr Stanczyk)
78	* Added explicit copy constructors to Imath::Matrix33<T> and
79	  ImathMatrix44<T> to make conversions between float and double
80	  matrices more convenient.
81	  (Florian Kainz)
82	* Added slerpShortestArc() and euclideanInnerProduct() functions
83	  to Imath::Quat<T>.
84	  (Nick Porcino)
85	* Added 4D vector class template Imath::Vec4<T>.
86	  (Nick Porcino)
87	* Copy constructors and assignment operators for Matrix33<T>
88	  and Matrix44<T> are up to 25% faster.  Added matrix constructors
89	  that do not initialize the matrix (this is faster in cases where
90	  the initial value of the matrix is immediately overwritten anyway).
91	  (Nick Porcino)
92	* Rewrote function closestPointOnBox(point,box).  Shortened
93	  the code, improved numerical accuracy, fixed a bug where
94	  closestPointOnBox(box.center(),box) would return the center
95	  of the +Z side of the box, even if the +/-X or +/-Y sides
96	  were closer.
97	  (Florian Kainz)
98	* Rewrote function findEntryAndExitPoints() in ImathBoxAlgo.h.
99	  Results are now consistent with those from intersect(), also
100	  in ImathBoxAlgo.h.
101	  (Florian Kainz)
102	* Made Vec2<T>::length() and Vec3<T>::length() more accurate for
103	  vectors whose length is less than sqrt(limits<T>::smallest());
104	  (Florian Kainz)
105	* Made Quat<T>::angle() more accurate for small angles.
106	  (Don Hatch)
107
108Version 1.0.1:
109	* Removed Windows .suo files from distribution.
110	  (Eric Wimmer)
111
112Version 1.0.0:
113	* Bumped DSO version number to 6.0
114	  (Florian Kainz)
115	* Rounding during float-to-half conversion now implements
116	  "round to nearest even" mode: if the original float value
117	  is exactly in the middle between the two closest half values
118	  then rounding chooses the half value whose least significant
119	  bit is zero.
120	  (Florian Kainz)
121	* Installation Tuning:
122	  - Corrected version number on dso's (libtool) - now 5.0
123	  - Separated ILMBASE_LDFLAGS and ILMBASE_LIBS so that test programs
124	    in configure scripts of packages dependent on IlmBase can link
125	    with static libraries properly
126	  - eliminated some warning messages during install
127	  (Andrew Kunz)
128
129Version 0.9.0:
130	* Initial release of this code as a separate library.
131	  Previously the libraries contained were part of
132	  version 1.4.0 of OpenEXR
133	* New build scripts for Linux/Unix
134	  (Andrew Kunz)
135	* New Windows project files and build scripts
136	  (Kimball Thurston)
137