ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/liblzf/Changes
Revision: 1.55
Committed: Thu Feb 16 05:43:19 2012 UTC (12 years, 3 months ago) by root
Branch: MAIN
Changes since 1.54: +3 -0 lines
Log Message:
*** empty log message ***

File Contents

# Content
1
2 TODO: try unaligned copy again in decompressor
3 TODO: allow size-optimised binaries by avoiding unrolling
4
5 - switch to a multiplicative hash (developed with Steinar Gunderson),
6 which is faster on modern cpus and compresses a bit better. The old
7 hash function which uses only shifts is still available.
8 - allow user configurable hash table slots, which makes it possible
9 to use e.g. 16 bit offsets for a smaller hashtable (if your data is
10 always < 64kb).
11 - use _WIN32, not WIN32, when testing for windows (fails with bcc),
12 patch by Tamas Tevesz.
13
14 3.6 Mon Feb 7 17:37:31 CET 2011
15 - fixed hash calculation in C♯ version (Tiago Freitas Leal).
16 - unroll copy for small sizes, use memcpy for larger sizes,
17 greatly speeding up decompression in most cases.
18 - finally disable rep movsb - it's a big loss on modern intel cpus,
19 and only a small win on amd cpus.
20 - improve C++ compatibility of the code.
21 - slightly improve compressor speed.
22 - halved memory requirements for compressor on 64 bit architectures,
23 which can improve the speed quite a bit on older cpus.
24
25 3.5 Fri May 1 02:28:42 CEST 2009
26 - lzf_compress did sometimes write one octet past the given output
27 buffer (analyzed and nice testcase by Salvatore Sanfilippo).
28
29 3.4 Tue Sep 2 06:45:00 CEST 2008
30 - the fix from 3.3 introduced a compression bug, which is fixed in
31 this release (which explains the mysterious prerelease...). Thanks
32 once more to Clément Calmels.
33
34 3.3 Mon Aug 25 03:17:42 CEST 2008
35 - lzf_compress could access memory after the given input buffer
36 when outputting back references. reported with nice testcase
37 by Clément Calmels.
38
39 3.2 Fri May 9 18:52:23 CEST 2008
40 - include a workaround for failing POSIX and real-world compliance
41 on 64 bit windows (microsoft claims to support POSIX, but is far
42 from it). (bug found and analysed nicely by John Lilley).
43
44 3.1 Fri Nov 30 11:33:04 CET 2007
45 - IMPORTANT BUGFIX: a too long final literal run would corrupt data
46 in the encoder (this was introduced in 3.0 only, earlier versions
47 are safe).
48
49 3.0 Tue Nov 13 22:13:09 CET 2007
50 - switched to 2-clause bsd with "GPL v2 or any later version" option.
51 - speed up compression by ~10-15% in common cases
52 by some manual unrolling.
53 - import some compiler tricks from JSON::XS, for further speed-ups.
54 - tune hash functions depending on ULTRA_FAST or VERY_FAST settings.
55 - for typical binary data (e.g. /bin/bash, memory dumps,
56 canterbury corpus etc.), speed is now comparable to fastlz, but
57 with better compression ratio. with ULTRA_FAST, it's typically
58 3-15% faster than fastlz while still maintaining a similar ratio.
59 (amd64 and core 2 duo, ymmv). thanks a lot for the competition :)
60 - undo inline assembly in compressor, it is no longer helpful.
61 - no changes to the decompressor.
62 - use a HLOG of 16 by default now (formerly 15).
63
64 2.1 Fri Nov 2 13:34:42 CET 2007
65 - switched to a 2-clause bsd license with GPL exception.
66 - get rid of memcpy.
67 - tentatively use rep movsb on x86 and x86_64 (gcc only) for a
68 moderate speed improvement.
69 - applied patch by Kein-Hong Man to maske lzf.c compile under
70 the crippled mingw32 environment.
71
72 2.0 Fri Feb 16 23:11:18 CET 2007
73 - replaced lzf demo by industrial-strength lzf utility with behaviour
74 similar other compression utilities. Thanks for Stefan Traby for
75 rewriting it!
76 - fix state arg prototype.
77
78 1.7 Wed Sep 27 17:29:15 CEST 2006
79 - remove bogus "unlzf" patch.
80 note to self: never accept well-meant patches.
81 - make lzf more robust in presence of padding bytes or sudden eof.
82
83 1.6 Fri Jul 7 17:31:26 CEST 2006
84 - the lzf example utility will now uncompress if invoked
85 as "unlzf" (patch by Scott Feeney).
86 - add CHECK_INPUT option that adds more checks for input
87 data validity.
88 - help applications that do not pass in the correct length
89 (such as php) by returning either EINVAL or E2BIG.
90 - default HLOG size is now 15 (cpu caches have increased).
91 - documentation fixes.
92
93 1.51 Thu Apr 14 22:15:46 CEST 2005
94 - incorporated C♯ implementation of both the en- and decoder,
95 written by "Oren J. Maurice".
96 You can find it in the cs/ subdirectory.
97 - make FRST, NEXT IDX overridable if lzf_c.c is directly included
98 in the code.
99
100 1.5 Tue Mar 8 20:23:23 CET 2005
101 - incorporated improvements by Adam D. Moss,
102 which includes a new VERY_FAST mode which is
103 a bit slower than ULTRA_FAST but much better,
104 and enabled it as default.
105
106 1.401 Thu Mar 3 18:00:52 CET 2005
107 - use cstring in c++, not string.h.
108 - change of contact address.
109
110 1.4 Wed Dec 15 08:08:49 CET 2004
111 - very very slight tuning of the hashing function.
112
113 1.3 Thu Mar 25 15:41:17 CET 2004
114 - changed license of lzf core code to explicitly allow
115 relicensing under the GPLv2.
116 - added VPATH support as suggested by Björn Eriksson.
117
118 1.2 Mon Dec 29 13:47:28 CET 2003
119 - avoid spurious memory accesses after the to-be-compressed
120 memory region. originally reported by Michal Zalewski.
121 - flip LZF_STACK_ARG meaning (to be correct).
122
123 1.1 Tue Dec 23 05:48:32 CET 2003
124 - removed #warn directive, it's not worth the hassle.
125 - add LZF_STACK_ARG and AVOID_ERRNO configurations
126 for embedded systems.
127 - make it compile cleanly as c++.
128 - some small documentation and code fixes.
129
130 1.0 Sun Nov 17 12:37:37 CET 2002
131 - slightly better compression ratio, almost unmeasurably
132 slower.
133 - some documentation fixes.
134
135 0.4 Thu Jun 13 14:11:10 CEST 2002
136 - typoe fix.
137 - lzf demo program now properly decompresses small files.
138 - fix another 64 bit issue, found by Laurent Deniel.
139
140 0.3 Tue Jan 16 13:21:14 CET 2001
141 - fix silly beginners 32/64 bit mistake.
142
143 0.2 Thu Jan 4 05:56:42 CET 2001
144 - now totally independent of autoconfig, for
145 easy inclusion into other programs.
146 - much better fine-tuning, faster and better than 0.1.
147
148 0.1 2000
149 - initial release.