13.6  Mon Feb  7 17:37:31 CET 2011
2	- fixed hash calculation in C♯ version (Tiago Freitas Leal).
3        - unroll copy for small sizes, use memcpy for larger sizes,
4          greatly speeding up decompression in most cases.
5        - finally disable rep movsb - it's a big loss on modern intel cpus,
6          and only a small win on amd cpus.
7        - improve C++ compatibility of the code.
8        - slightly improve compressor speed.
9        - halved memory requirements for compressor on 64 bit architectures,
10          which can improve the speed quite a bit on older cpus.
11
123.5  Fri May  1 02:28:42 CEST 2009
13	- lzf_compress did sometimes write one octet past the given output
14          buffer (analyzed and nice testcase by Salvatore Sanfilippo).
15
163.4  Tue Sep  2 06:45:00 CEST 2008
17	- the fix from 3.3 introduced a compression bug, which is fixed in
18          this release (which explains the mysterious prerelease...). Thanks
19          once more to Clément Calmels.
20
213.3  Mon Aug 25 03:17:42 CEST 2008
22	- lzf_compress could access memory after the given input buffer
23          when outputting back references. reported with nice testcase
24          by Clément Calmels.
25
263.2  Fri May  9 18:52:23 CEST 2008
27	- include a workaround for failing POSIX and real-world compliance
28          on 64 bit windows (microsoft claims to support POSIX, but is far
29          from it). (bug found and analysed nicely by John Lilley).
30
313.1  Fri Nov 30 11:33:04 CET 2007
32	- IMPORTANT BUGFIX: a too long final literal run would corrupt data
33          in the encoder (this was introduced in 3.0 only, earlier versions
34          are safe).
35
363.0  Tue Nov 13 22:13:09 CET 2007
37	- switched to 2-clause bsd with "GPL v2 or any later version" option.
38        - speed up compression by ~10-15% in common cases
39          by some manual unrolling.
40        - import some compiler tricks from JSON::XS, for further speed-ups.
41        - tune hash functions depending on ULTRA_FAST or VERY_FAST settings.
42        - for typical binary data (e.g. /bin/bash, memory dumps,
43          canterbury corpus etc.), speed is now comparable to fastlz, but
44          with better compression ratio. with ULTRA_FAST, it's typically
45          3-15% faster than fastlz while still maintaining a similar ratio.
46          (amd64 and core 2 duo, ymmv). thanks a lot for the competition :)
47        - undo inline assembly in compressor, it is no longer helpful.
48        - no changes to the decompressor.
49        - use a HLOG of 16 by default now (formerly 15).
50
512.1  Fri Nov  2 13:34:42 CET 2007
52	- switched to a 2-clause bsd license with GPL exception.
53	- get rid of memcpy.
54        - tentatively use rep movsb on x86 and x86_64 (gcc only) for a
55          moderate speed improvement.
56        - applied patch by Kein-Hong Man to maske lzf.c compile under
57          the crippled mingw32 environment.
58
592.0  Fri Feb 16 23:11:18 CET 2007
60	- replaced lzf demo by industrial-strength lzf utility with behaviour
61          similar other compression utilities. Thanks for Stefan Traby for
62          rewriting it!
63        - fix state arg prototype.
64
651.7  Wed Sep 27 17:29:15 CEST 2006
66	- remove bogus "unlzf" patch.
67          note to self: never accept well-meant patches.
68	- make lzf more robust in presence of padding bytes or sudden eof.
69
701.6  Fri Jul  7 17:31:26 CEST 2006
71	- the lzf example utility will now uncompress if invoked
72          as "unlzf" (patch by Scott Feeney).
73        - add CHECK_INPUT option that adds more checks for input
74          data validity.
75        - help applications that do not pass in the correct length
76          (such as php) by returning either EINVAL or E2BIG.
77        - default HLOG size is now 15 (cpu caches have increased).
78        - documentation fixes.
79
801.51 Thu Apr 14 22:15:46 CEST 2005
81	- incorporated C♯ implementation of both the en- and decoder,
82          written by "Oren J. Maurice".
83          You can find it in the cs/ subdirectory.
84        - make FRST, NEXT IDX overridable if lzf_c.c is directly included
85          in the code.
86
871.5  Tue Mar  8 20:23:23 CET 2005
88	- incorporated improvements by Adam D. Moss,
89          which includes a new VERY_FAST mode which is
90          a bit slower than ULTRA_FAST but much better,
91          and enabled it as default.
92
931.401 Thu Mar  3 18:00:52 CET 2005
94	- use cstring in c++, not string.h.
95        - change of contact address.
96
971.4  Wed Dec 15 08:08:49 CET 2004
98	- very very slight tuning of the hashing function.
99
1001.3  Thu Mar 25 15:41:17 CET 2004
101	- changed license of lzf core code to explicitly allow
102          relicensing under the GPLv2.
103	- added VPATH support as suggested by Björn Eriksson.
104
1051.2  Mon Dec 29 13:47:28 CET 2003
106	- avoid spurious memory accesses after the to-be-compressed
107          memory region. originally reported by Michal Zalewski.
108        - flip LZF_STACK_ARG meaning (to be correct).
109
1101.1  Tue Dec 23 05:48:32 CET 2003
111	- removed #warn directive, it's not worth the hassle.
112        - add LZF_STACK_ARG and AVOID_ERRNO configurations
113          for embedded systems.
114        - make it compile cleanly as c++.
115        - some small documentation and code fixes.
116
1171.0  Sun Nov 17 12:37:37 CET 2002
118	- slightly better compression ratio, almost unmeasurably
119          slower.
120        - some documentation fixes.
121
1220.4  Thu Jun 13 14:11:10 CEST 2002
123	- typoe fix.
124        - lzf demo program now properly decompresses small files.
125        - fix another 64 bit issue, found by Laurent Deniel.
126
1270.3  Tue Jan 16 13:21:14 CET 2001
128	- fix silly beginners 32/64 bit mistake.
129
1300.2  Thu Jan  4 05:56:42 CET 2001
131	- now totally independent of autoconfig, for
132          easy inclusion into other programs.
133        - much better fine-tuning, faster and better than 0.1.
134
1350.1  2000
136	- initial release.
137