I have question for David but I want to make it public:
Are there cases when BWTS could provide significantly better compression that ordinary BWT? I am wondering if BWTS could help in case of eg sorted dictionary. On http://www.maximumcompression.com/data/dict.php BWT based (no STx based or LZP augmented) compressors perform poorly. What is the average number of Lyndon words after factorization?
Additionally I think that in paper: http://bijective.dogma.net/00yyy.pdf on page 6, Algorithm 3.2 Match(n) there is error in line 8. Currently loop at lines 7-8 does nothing as array before contains only zeroes. I think in line 8 should be: before[c] = before[c - 1] + counts[c - 1]