FIX to Practical Indexing of Repetitive Collections using Relative Lempel-Ziv

Level: Serious

When preparing the journal version, I discovered that the experimental results were all wrong. Both space and time are higher than what appears in the paper, making our scheme lose in practice to the Hybrid Index. Another mistake is in line 3 of page 5, where it says that we consider the whole prefix S_k[1..p_u-1]. If we index the whole prefixes, we might report duplicate occurrences. We must consider instead only the block, S_k[p_{u-1}..p_u-1].