ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/liblzf/Changes
Revision: 1.57
Committed: Fri Mar 30 20:58:09 2012 UTC (12 years, 1 month ago) by root
Branch: MAIN
Changes since 1.56: +3 -1 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 TODO: implement lzf_c_best in lzf.
5 TODO: fix lzf_c_best.
6
7 - switch to a multiplicative hash (developed with Steinar Gunderson),
8 which is faster on modern cpus and compresses a bit better. The old
9 hash function which uses only shifts is still available.
10 - allow user configurable hash table slots, which makes it possible
11 to use e.g. 16 bit offsets for a smaller hashtable (if your data is
12 always < 64kb).
13 - use _WIN32, not WIN32, when testing for windows (fails with bcc),
14 patch by Tamas Tevesz.
15 - try to port to win64+gcc, which is non-posix but doesn't have _int64
16 (reported by Maciej Adamczyk).
17
18 3.6 Mon Feb 7 17:37:31 CET 2011
19 - fixed hash calculation in C♯ version (Tiago Freitas Leal).
20 - unroll copy for small sizes, use memcpy for larger sizes,
21 greatly speeding up decompression in most cases.
22 - finally disable rep movsb - it's a big loss on modern intel cpus,
23 and only a small win on amd cpus.
24 - improve C++ compatibility of the code.
25 - slightly improve compressor speed.
26 - halved memory requirements for compressor on 64 bit architectures,
27 which can improve the speed quite a bit on older cpus.
28
29 3.5 Fri May 1 02:28:42 CEST 2009
30 - lzf_compress did sometimes write one octet past the given output
31 buffer (analyzed and nice testcase by Salvatore Sanfilippo).
32
33 3.4 Tue Sep 2 06:45:00 CEST 2008
34 - the fix from 3.3 introduced a compression bug, which is fixed in
35 this release (which explains the mysterious prerelease...). Thanks
36 once more to Clément Calmels.
37
38 3.3 Mon Aug 25 03:17:42 CEST 2008
39 - lzf_compress could access memory after the given input buffer
40 when outputting back references. reported with nice testcase
41 by Clément Calmels.
42
43 3.2 Fri May 9 18:52:23 CEST 2008
44 - include a workaround for failing POSIX and real-world compliance
45 on 64 bit windows (microsoft claims to support POSIX, but is far
46 from it). (bug found and analysed nicely by John Lilley).
47
48 3.1 Fri Nov 30 11:33:04 CET 2007
49 - IMPORTANT BUGFIX: a too long final literal run would corrupt data
50 in the encoder (this was introduced in 3.0 only, earlier versions
51 are safe).
52
53 3.0 Tue Nov 13 22:13:09 CET 2007
54 - switched to 2-clause bsd with "GPL v2 or any later version" option.
55 - speed up compression by ~10-15% in common cases
56 by some manual unrolling.
57 - import some compiler tricks from JSON::XS, for further speed-ups.
58 - tune hash functions depending on ULTRA_FAST or VERY_FAST settings.
59 - for typical binary data (e.g. /bin/bash, memory dumps,
60 canterbury corpus etc.), speed is now comparable to fastlz, but
61 with better compression ratio. with ULTRA_FAST, it's typically
62 3-15% faster than fastlz while still maintaining a similar ratio.
63 (amd64 and core 2 duo, ymmv). thanks a lot for the competition :)
64 - undo inline assembly in compressor, it is no longer helpful.
65 - no changes to the decompressor.
66 - use a HLOG of 16 by default now (formerly 15).
67
68 2.1 Fri Nov 2 13:34:42 CET 2007
69 - switched to a 2-clause bsd license with GPL exception.
70 - get rid of memcpy.
71 - tentatively use rep movsb on x86 and x86_64 (gcc only) for a
72 moderate speed improvement.
73 - applied patch by Kein-Hong Man to maske lzf.c compile under
74 the crippled mingw32 environment.
75
76 2.0 Fri Feb 16 23:11:18 CET 2007
77 - replaced lzf demo by industrial-strength lzf utility with behaviour
78 similar other compression utilities. Thanks for Stefan Traby for
79 rewriting it!
80 - fix state arg prototype.
81
82 1.7 Wed Sep 27 17:29:15 CEST 2006
83 - remove bogus "unlzf" patch.
84 note to self: never accept well-meant patches.
85 - make lzf more robust in presence of padding bytes or sudden eof.
86
87 1.6 Fri Jul 7 17:31:26 CEST 2006
88 - the lzf example utility will now uncompress if invoked
89 as "unlzf" (patch by Scott Feeney).
90 - add CHECK_INPUT option that adds more checks for input
91 data validity.
92 - help applications that do not pass in the correct length
93 (such as php) by returning either EINVAL or E2BIG.
94 - default HLOG size is now 15 (cpu caches have increased).
95 - documentation fixes.
96
97 1.51 Thu Apr 14 22:15:46 CEST 2005
98 - incorporated C♯ implementation of both the en- and decoder,
99 written by "Oren J. Maurice".
100 You can find it in the cs/ subdirectory.
101 - make FRST, NEXT IDX overridable if lzf_c.c is directly included
102 in the code.
103
104 1.5 Tue Mar 8 20:23:23 CET 2005
105 - incorporated improvements by Adam D. Moss,
106 which includes a new VERY_FAST mode which is
107 a bit slower than ULTRA_FAST but much better,
108 and enabled it as default.
109
110 1.401 Thu Mar 3 18:00:52 CET 2005
111 - use cstring in c++, not string.h.
112 - change of contact address.
113
114 1.4 Wed Dec 15 08:08:49 CET 2004
115 - very very slight tuning of the hashing function.
116
117 1.3 Thu Mar 25 15:41:17 CET 2004
118 - changed license of lzf core code to explicitly allow
119 relicensing under the GPLv2.
120 - added VPATH support as suggested by Björn Eriksson.
121
122 1.2 Mon Dec 29 13:47:28 CET 2003
123 - avoid spurious memory accesses after the to-be-compressed
124 memory region. originally reported by Michal Zalewski.
125 - flip LZF_STACK_ARG meaning (to be correct).
126
127 1.1 Tue Dec 23 05:48:32 CET 2003
128 - removed #warn directive, it's not worth the hassle.
129 - add LZF_STACK_ARG and AVOID_ERRNO configurations
130 for embedded systems.
131 - make it compile cleanly as c++.
132 - some small documentation and code fixes.
133
134 1.0 Sun Nov 17 12:37:37 CET 2002
135 - slightly better compression ratio, almost unmeasurably
136 slower.
137 - some documentation fixes.
138
139 0.4 Thu Jun 13 14:11:10 CEST 2002
140 - typoe fix.
141 - lzf demo program now properly decompresses small files.
142 - fix another 64 bit issue, found by Laurent Deniel.
143
144 0.3 Tue Jan 16 13:21:14 CET 2001
145 - fix silly beginners 32/64 bit mistake.
146
147 0.2 Thu Jan 4 05:56:42 CET 2001
148 - now totally independent of autoconfig, for
149 easy inclusion into other programs.
150 - much better fine-tuning, faster and better than 0.1.
151
152 0.1 2000
153 - initial release.