ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/liblzf/Changes
(Generate patch)

Comparing liblzf/Changes (file contents):
Revision 1.14 by root, Thu Mar 3 17:08:17 2005 UTC vs.
Revision 1.53 by root, Fri Mar 4 16:42:20 2011 UTC

1
2 - use _WIN32, not WIN32, when testing for windows (fails with bcc),
3 patch by Tamas Tevesz.
4
53.6 Mon Feb 7 17:37:31 CET 2011
6 - fixed hash calculation in C♯ version (Tiago Freitas Leal).
7 - unroll copy for small sizes, use memcpy for larger sizes,
8 greatly speeding up decompression in most cases.
9 - finally disable rep movsb - it's a big loss on modern intel cpus,
10 and only a small win on amd cpus.
11 - improve C++ compatibility of the code.
12 - slightly improve compressor speed.
13 - halved memory requirements for compressor on 64 bit architectures,
14 which can improve the speed quite a bit on older cpus.
15
163.5 Fri May 1 02:28:42 CEST 2009
17 - lzf_compress did sometimes write one octet past the given output
18 buffer (analyzed and nice testcase by Salvatore Sanfilippo).
19
203.4 Tue Sep 2 06:45:00 CEST 2008
21 - the fix from 3.3 introduced a compression bug, which is fixed in
22 this release (which explains the mysterious prerelease...). Thanks
23 once more to Clément Calmels.
24
253.3 Mon Aug 25 03:17:42 CEST 2008
26 - lzf_compress could access memory after the given input buffer
27 when outputting back references. reported with nice testcase
28 by Clément Calmels.
29
303.2 Fri May 9 18:52:23 CEST 2008
31 - include a workaround for failing POSIX and real-world compliance
32 on 64 bit windows (microsoft claims to support POSIX, but is far
33 from it). (bug found and analysed nicely by John Lilley).
34
353.1 Fri Nov 30 11:33:04 CET 2007
36 - IMPORTANT BUGFIX: a too long final literal run would corrupt data
37 in the encoder (this was introduced in 3.0 only, earlier versions
38 are safe).
39
403.0 Tue Nov 13 22:13:09 CET 2007
41 - switched to 2-clause bsd with "GPL v2 or any later version" option.
42 - speed up compression by ~10-15% in common cases
43 by some manual unrolling.
44 - import some compiler tricks from JSON::XS, for further speed-ups.
45 - tune hash functions depending on ULTRA_FAST or VERY_FAST settings.
46 - for typical binary data (e.g. /bin/bash, memory dumps,
47 canterbury corpus etc.), speed is now comparable to fastlz, but
48 with better compression ratio. with ULTRA_FAST, it's typically
49 3-15% faster than fastlz while still maintaining a similar ratio.
50 (amd64 and core 2 duo, ymmv). thanks a lot for the competition :)
51 - undo inline assembly in compressor, it is no longer helpful.
52 - no changes to the decompressor.
53 - use a HLOG of 16 by default now (formerly 15).
54
552.1 Fri Nov 2 13:34:42 CET 2007
56 - switched to a 2-clause bsd license with GPL exception.
57 - get rid of memcpy.
58 - tentatively use rep movsb on x86 and x86_64 (gcc only) for a
59 moderate speed improvement.
60 - applied patch by Kein-Hong Man to maske lzf.c compile under
61 the crippled mingw32 environment.
62
632.0 Fri Feb 16 23:11:18 CET 2007
64 - replaced lzf demo by industrial-strength lzf utility with behaviour
65 similar other compression utilities. Thanks for Stefan Traby for
66 rewriting it!
67 - fix state arg prototype.
68
691.7 Wed Sep 27 17:29:15 CEST 2006
70 - remove bogus "unlzf" patch.
71 note to self: never accept well-meant patches.
72 - make lzf more robust in presence of padding bytes or sudden eof.
73
741.6 Fri Jul 7 17:31:26 CEST 2006
75 - the lzf example utility will now uncompress if invoked
76 as "unlzf" (patch by Scott Feeney).
77 - add CHECK_INPUT option that adds more checks for input
78 data validity.
79 - help applications that do not pass in the correct length
80 (such as php) by returning either EINVAL or E2BIG.
81 - default HLOG size is now 15 (cpu caches have increased).
82 - documentation fixes.
83
841.51 Thu Apr 14 22:15:46 CEST 2005
85 - incorporated C♯ implementation of both the en- and decoder,
86 written by "Oren J. Maurice".
87 You can find it in the cs/ subdirectory.
88 - make FRST, NEXT IDX overridable if lzf_c.c is directly included
89 in the code.
90
911.5 Tue Mar 8 20:23:23 CET 2005
92 - incorporated improvements by Adam D. Moss,
93 which includes a new VERY_FAST mode which is
94 a bit slower than ULTRA_FAST but much better,
95 and enabled it as default.
1 96
21.401 Thu Mar 3 18:00:52 CET 2005 971.401 Thu Mar 3 18:00:52 CET 2005
3 - use cstring in c++, not string.h. 98 - use cstring in c++, not string.h.
4 - change of contact address. 99 - change of contact address.
5 100

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines