An algorithm to produce a sequence of insert and copy commands (an edit script) which creates a new version file from a reference file. To begin, hash every block of the reference file and store every hash in a hash value array. Build a suffix array and three other data structures for quick access. Beginning at the first location in the version file, hash a block and look for the longest match in the reference file. Upon a match, encode an insert back to the previous match and a copy of the match. If no match, look at the next location. At the end, encode an insert for remaining unmatched characters.