Age | Commit message (Collapse) | Author | |
---|---|---|---|
2007-12-17 | Call main_input() recursively for main_merge_arguments. | josh.macdonald | |
2007-12-15 | Compile on OS X, better implementation of UNALIGNED_OK=0 small checksum. | josh.macdonald | |
2007-12-14 | Fix a few more uint uses. | josh.macdonald | |
2007-12-14 | Replace uint with usize_t, more portable. | josh.macdonald | |
2007-12-14 | Implement UNALIGNED_OK for xd3_small_cksum_update, hoping to fix | josh.macdonald | |
test failure on Xscale/ARM platform. | |||
2007-12-13 | A few cleanups -- xd3_scksum should have been using uint32_t, but I | josh.macdonald | |
doubt that is the real problem. I suspect the problem on ARM/Xscale has to do with unaligned access issues. | |||
2007-12-07 | Patches from A. Mennucc for Debian. Fixing makfile for Cygwin | josh.macdonald | |
conditional compilation, which is irregular. | |||
2007-12-06 | Adjust the low-end range for command_line_arguments unittest. I wish this | josh.macdonald | |
didn't depend on the system rand() function, but it's portable at least. | |||
2007-12-05 | Fix bug in xd3_set_source, the optimization for shiftby/maskby were | josh.macdonald | |
in some cases not setting those variables, which are used by xd3_blksize_div | |||
2007-12-02 | Fix recode, recode test. | josh.macdonald | |
2007-12-02 | Working on the test failure in dbeug mode, partial recode test written. | josh.macdonald | |
2007-11-27 | Diagnose expected non-failures test failure introfuced in 3.0s relating | josh.macdonald | |
to DJW and compiler optimization level. Mysterious. | |||
2007-11-26 | The rabin-karp checksum looks better in testing but doesn't really | josh.macdonald | |
seem to improve things in practice. Removed HASH_PRIME. | |||
2007-11-13 | (no commit message) | josh.macdonald | |
2007-11-11 | Compile with g++ 3.4.4 and add C++ checksum_test.cc | josh.macdonald | |
2007-11-10 | Adds output regarding inefficient copy instructions to "printdelta". | josh.macdonald | |
Adds crude inefficiency check to xd3_smatch(), preventing 4-byte matches if (here-addr) >= 2^14 and 5-byte matches if (here-addr) >= 2^21. Nice improvement. | |||
2007-11-09 | Release 3.0s | josh.macdonald | |
2007-11-07 | A few optimizations: -1 (fastest) is faster than before. -fno-builtin | josh.macdonald | |
makes a difference. New implementation of xd3_forward_match() is written and tested, but not enabled, needs a little more benchmarking. | |||
2007-10-29 | Add a new COMPLEVEL_1, previous -1 becomes -2. | josh.macdonald | |
2007-10-29 | Add a new fastest, fastest->faster (-2) | josh.macdonald | |
2007-10-29 | Inline some functions. | josh.macdonald | |
2007-10-29 | Use maskby/shiftby when source->size <= source->blksize, which helps | josh.macdonald | |
xd3_process_memory() uses. | |||
2007-10-28 | New xd3_blksize_div() function used everywhere to avoid __umoddi3 | josh.macdonald | |
2007-10-28 | Eliminate calls to __umoddi3 (compiler generated for 64-bit % 32-bit). | josh.macdonald | |
More 80col reformatting. | |||
2007-10-28 | 80col changes. | josh.macdonald | |
2007-10-28 | ENC_FLUSH -> ENC_INSTR | josh.macdonald | |
2007-10-28 | Reformat for 80 cols. Allow -S djwN where N is 0-9. Default | josh.macdonald | |
(for -S djw) is 3. -S djw9 enables multi-group huffman for all three sections. -S djw3 enables single-group huffman for all three sections. 1-2 are single data-only, 3-4 are single data+inst, 5-6 are single data+inst+addr, 7, 8, 9 enable multi-group huffman for the three sections in order. | |||
2007-10-28 | Comments. | josh.macdonald | |
2007-10-28 | Reformat >80col lines. 32-bit build/test fixes. | josh.macdonald | |
2007-10-21 | Comments and build with speed_test.cc | josh.macdonald | |
2007-10-20 | working on inline functions | josh.macdonald | |
2007-09-29 | Adds a test that triggers the assertion failure in djw_build_prefix, | josh.macdonald | |
a few cleanups, but does not fix the problem (next CL). | |||
2007-09-22 | Fixes: | josh.macdonald | |
1. Encoder was improperly setting stream->min_match on re-entry following XD3_GETSRCBLK events. 2. Decoder was improperly terminating a loop when XD3_GETSRCBLK events occur while decoding the final instruction. 3. An known assertion failure in djw_compute_mtf_1_2 will abort() in non-debug mode instead of running out of memory. | |||
2007-09-12 | Fixes xdelta recode command, but does not implement adler32, source | josh.macdonald | |
info, app header, etc. | |||
2007-09-12 | Add "xdelta3 recode" function, which is currently outputing incorrect | josh.macdonald | |
VCDIFF windows. | |||
2007-03-24 | Rename #define P to #define DP | josh.macdonald | |
2007-02-19 | cleanup regtest 400 test | josh.macdonald | |
2007-02-18 | pre 3.0p, all tests pass | josh.macdonald | |
2007-02-18 | remove PROMOTE, SSMATCH, TRYLAZY, raise default IOPT_SIZE | josh.macdonald | |
2007-02-18 | raise default sprevsz, add command-line flag -P to set | josh.macdonald | |
2007-02-14 | fix two alloc checks in djw.h | josh.macdonald | |
fix setting of stream->iopt_size fix small_page_test.c | |||
2007-02-14 | add small_page_test.c, currently fails | josh.macdonald | |
2007-02-13 | Adds xd3_flags for setting compression level in xd3_encode_memory(). | josh.macdonald | |
2007-02-12 | Fix and recomment xd3_srcwin_move_point() after changes in SVN 127 to use ↵ | josh.macdonald | |
src->blksize as the advance size. | |||
2007-02-11 | Fixes bugs in XDELTA environment variable. | josh.macdonald | |
Removes stream.srcwin_size variable, uses stream.src->blksize instead, to address performance regression from xdelta1. Changes the order of checksum computation to be like xdelta1, insert checksums in reverse to give preference to earlier matches. Removes svn:executable from files. | |||
2007-02-08 | Python swig module working with xd3_encode_memory(), xd3_decode_memory(), | josh.macdonald | |
and xd3_main_cmdline(). module added to xdelta3-regtest.py run-length experiments. XD3_INVALID_INPUT code added to distinguish decoder errors from invalid stream-states. xdelta3-test.py unittest added for python module. xd3_encode_memory() now limits config.winsize to <= 1MB: xd3_process_memory() now properly splits the input buffer into windows. | |||
2007-02-07 | Update FASTEST, DEFAULT compression level. Comments. | josh.macdonald | |
2007-02-04 | Adds a new compression level (-6), makes -1 faster and -9 better. | josh.macdonald | |
2007-02-04 | Decoder memory error, related to secondary compression, fixed. Valgrind | josh.macdonald | |
made this easy to spot. | |||
2007-02-01 | Implmeent xd3_encode_memory() and xd3_decode_memory(), add comments. | josh.macdonald | |