Home | History | Annotate | only in /src/external/bsd/zstd/dist/contrib/match_finders
Up to higher level directory
NameDateSize
README.md27-Oct-20242.2K
zstd_edist.c27-Oct-202419.8K
zstd_edist.h27-Oct-20243.2K

README.md

      1 ## Edit Distance Match Finder
      2 
      3 ```
      4 /* This match finder leverages techniques used in file comparison algorithms
      5  * to find matches between a dictionary and a source file.
      6  * 
      7  * The original motivation for studying this approach was to try and optimize 
      8  * Zstandard for the use case of patching: the most common scenario being 
      9  * updating an existing software package with the next version. When patching,
     10  * the difference between the old version of the package and the new version 
     11  * is generally tiny (most of the new file will be identical to 
     12  * the old one). In more technical terms, the edit distance (the minimal number 
     13  * of changes required to take one sequence of bytes to another) between the 
     14  * files would be small relative to the size of the file. 
     15  * 
     16  * Various 'diffing' algorithms utilize this notion of edit distance and 
     17  * the corresponding concept of a minimal edit script between two 
     18  * sequences to identify the regions within two files where they differ. 
     19  * The core algorithm used in this match finder is described in: 
     20  * 
     21  * "An O(ND) Difference Algorithm and its Variations", Eugene W. Myers,
     22  *    Algorithmica Vol. 1, 1986, pp. 251-266,
     23  *    <https://doi.org/10.1007/BF01840446>.
     24  * 
     25  * Additional algorithmic heuristics for speed improvement have also been included.
     26  * These we inspired from implementations of various regular and binary diffing 
     27  * algorithms such as GNU diff, bsdiff, and Xdelta. 
     28  * 
     29  * Note: after some experimentation, this approach proved to not provide enough 
     30  * utility to justify the additional CPU used in finding matches. The one area
     31  * where this approach consistently outperforms Zstandard even on level 19 is 
     32  * when compressing small files (<10 KB) using an equally small dictionary that 
     33  * is very similar to the source file. For the use case that this was intended,
     34  * (large similar files) this approach by itself took 5-10X longer than zstd-19 and 
     35  * generally resulted in 2-3X larger files. The core advantage that zstd-19 has 
     36  * over this approach for match finding is the overlapping matches. This approach 
     37  * cannot find any. 
     38  * 
     39  * I'm leaving this in the contrib section in case this ever becomes interesting 
     40  * to explore again.
     41  * */
     42 ```
     43