Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
* feat: IsByteZero works * fix: works on small test case * bench: awful * perf: use logderiv map * fix: "last byte" bug * style: remove redundant checks * fix: some bugs * feat: new data set and huffman estimations * chore: gitignore * chore: improved analytics * more data and an lzss compressor * feat: basic lzss decompressor * test: a couple * fix: compression works on the first 300b of calldata * fix: 1B addresses seem to work * test: some logging * fix: works on 2c2964. performance awful * perf: prefer actual backrefs for RLE; better but still bad * fix: works on 3c2943 with symb 0 * test: print compressed file size * feat: small tests work with indeterminate length * test: more state machine tests. failing * chore: comments/cleanup for lzss compression * fix: can handle two consecutive backrefs * test: more, failing "8zerosAfterNonzero" * test: zeroAfterNonzero added, fixed * test: twoZerosAfterNonzero * fix: zerosAfterNonzero pass * test: actually, 257zerosAfterNonzero fails * test: twoBackrefsAfterNonzero fails * fix: all "simple" tests pass * feat: basic i/o funcs * feat: preliminary snark decompressor impl * test: decompression snark * fix: write to the output table * fix: all zeros tests pass * fix: all tests pass, except for 3c2943: too slow * chore: remove prints * fix: bug with lone 0 in high indexes * test: add huffman estimated gains * perf: dfa search; actually makes things worse * revert: dfa search was counterproductive * refactor: compile 600KB * perf: ite -> api.Select * fix: bug with negative indexes * refac: remove useless functions * feat: some experiments with huffman coding * feat: bzip2 (bzip would be better) * fix: RLE bug * feat: huffman Decode * test: single symbol test * fix: two symbols test * revert: strange uppercase * bench: huffman decoding * perf: huffman improvement * refactor: lots of cleanup. one failing test remaining * fix: presumption of long negative space of zeros * bench: 26KB * update gitignore * fix: failed * safeguard * test: failing test for cs loading * bench: proving works * perf: reduce 1 lookup per backref * fix: read bugs * style: remove prints * perf: a few petty opts * perf: a few little opts * perf: custom constraint for inIDelta * perf: custom constraint for advancing inI * perf: custom constraint for copying * feat: api.IsNonZero * chore: clean up comments and prints * fix: trailing backref bug * refactor: massive simplification of lzssv1 compression * refactor: remove log heads * refactor: cleaning tests * feat: some analytics * fix: make builder private again * chore: better logging, remove code from data folder * fix remove extra zero from logs * make staticcheck happy(er) * feat: more analytics * skip testCompareCoverage * perf: "start at" * bench: don't inflate the decompressed size too much * test: added average batch test case * test: added BenchmarkAverageBatch * perf: make compress way faster * test: reactivate other for cmp * style: costmetics * checkpoint * fix perf: longer backref starting with runs of 0, but slower * some simplification * revert: remove TestCompressWithContext * revert: uncrowd the pr * chore: uncrowd the pr a bit more * fix: make tests pass * test skip calldata test temporarily * feat: pack/unpack functions * feat: bit-level alignment of compressed * test: compress_tests pass * test: pack * test: decompression works; must go about packing differently * feat: r/w num, (un)marshal for stream * test: marshal test * test: compression roundtrip passes w merged stream utils * refactor: simplify packing * revert overcomplicated packing * fix: snark errors * chore make staticcheck happy * test: average batch * fix: 18b offset - ave * checkpoint * perf: binary search of longest backref * feat: started v2 * perf: one binary search only * perf: add lazy match look ahead 1 * style: code cleaning * perf: even better lookahead for lazy deflate * perf: adjustement * checkpoint * perf: smaller backrefs * perf: faster compression by reducing search space * perf: reduce mem alloc when init suffix array * perf: replace sort.Search * fix: fuzzer bug * test: add testdata/ fuzzer dir * perf: don't use 0 as symbol delimiter * ensure symbol is in dict and reachable * feat: add multi symbol * check against savings * fix bug * feat: new stream * fix: small packing test works * refactor: packing as stream feature * bench: large * refactor: no need for outAt * remove duplicate bitio dep * feat: read lengths! * feat: offset, length and bytes tables * feat: snark decompressor, all but eof logic done * feat: snark decomp done, not yet tested * fix: ReadIntoStream * fix: some minor bugs * test: trying to recreate the length bug * fix: test with backrefs * test: 253-254-255 fails * fix: simple table lookup works * test: actual calldata * test: fuzz * refactor: get rid of lzssv1 * feat: compile large circuit outside tests * make sure all backref address lengths are multiples of 4 * fix: readIntoStream bug * feat: forceDivisibleBy * chore: minor changes to benchmark * chore: set word size to 1 * chore: remove unused code * chore: remove prints and all huffman code * refactor: lzssv2 -> lzss * refac: compression modes * chore: clean up test cases * chore: go.sum * revert: map.keys/values to be private as before * revert extra gitignore rules * docs: BestCompression vs BestSnarkDecomposition * minor fixes and removing prints * build: get gopter * fix: comment out go generate in suffixarray * fix: actually remove the go generate line * style: code cleaning * test: updates fuzz test with new API * fix ineffassign * test: update FuzzSnark * refactor: move _testing.go into _test.go file * feat: cache lookup blueprint entries in solving phase * clean: remove unused IsNonZero * refactor: remove hintwithname * refactor: remove gzwrite and gcd * fix: NewCombination when at least one var is const * fix: in test engine, commit should not return 0 with tiny field * refactor: NewCombination -> EvaluatePlonkExpression * fix: close zip writer * style: remove some dead code * fix: fixes #918 incorrect hash Sum usage * style: costmetic changes --------- Co-authored-by: Gautam Botrel <[email protected]>
- Loading branch information
