Commit graph

20 commits

Author SHA1 Message Date
devinacker
ab1823487b fix(?) a huge concurrency issue
(why the fuck did i think having any sort of global state in a
compression routine was a good idea?)
2014-09-28 23:45:48 -04:00
devinacker
e45f993eed don't check one more offset than we need
(another thing i forgot to update after changing the tuple size)
2014-04-03 18:15:45 -04:00
devinacker
7aa58df92f fix comment in compress.c 2014-01-13 16:25:01 -05:00
devinacker
d7ea381746 index 4 bytes instead of 3
(since 4 is the smallest back reference anyway; this doesn't have nearly
the same effect on speed as 2-3 did though)
2013-12-25 14:53:16 -05:00
devinacker
a3b9b3a5ef don't LZ search if an RLE is already max length 2013-11-15 17:09:54 -05:00
devinacker
82a1849337 forgot to actually free() the hash table (oops) 2013-11-15 16:52:18 -05:00
devinacker
9d52f2a03e add license text to opening comments
(just for those who want to copy compress.c/.h into their own projects)
2013-11-15 16:40:53 -05:00
devinacker
dd67967f35 more compression optimization (probably final)
using uthash to keep a hashtable of byte triples instead of an array of
byte pairs (since the minimum back reference size is 3 bytes)
2013-11-15 16:32:14 -05:00
devinacker
b1263f2c05 fix bad decompression size error 2013-11-15 16:23:10 -05:00
devinacker
fb163a5b08 more consistent types, some bounds checking 2013-11-13 22:31:06 -05:00
Anthony J. Bentley
dd17790576 Fix printf format warning. Remove leading space from positive value. 2013-11-12 03:04:57 -07:00
devinacker
66233bc3a9 s->t 2013-10-29 08:47:27 -04:00
devinacker
f8ed01c458 use copy loops instead of memcpy/memmove
because different versions of gcc seem to disagree on how both of these
functions should handle overlapping memory. thanks stallman
2013-10-26 14:41:21 -04:00
devinacker
e8fb851b50 fix up compress.h
so that it can be used with c++ more conveniently
2013-10-26 14:25:23 -04:00
devinacker
d39f07390e fix small bug w/ 16-bit RLE, adjust some printfs 2013-10-24 17:55:41 -04:00
devinacker
cf3dda681f massive LZ search optimization (needs tweaking)
output is now a couple of bytes larger sometimes for some reason but
seems to decompress fine, and the compression is 3-4x faster which is
great
2013-10-24 01:08:16 -04:00
devinacker
23c4b03c56 remove some old stupid code 2013-10-23 10:51:58 -04:00
devinacker
11a23403bc misc cleanup for real release 2013-10-22 17:02:37 -04:00
Devin Acker
c9e004e18b add "fast" compression mode 2013-06-05 19:50:07 -04:00
Devin Acker
23a609086f this is where I try out the github windows client
Ten years ago we had Steve Jobs, Johnny Cash, and Bob Hope. Now, we have
no jobs, no cash, and no hope.
2013-06-04 21:56:07 -04:00