The current comparison algorithm is very similar to the C implementation, which was helpful during the re-implementation. However, it's very hard to read and follow. The current code documentation also reflects the complexity of the comparison algorithm very poorly. It's however very important for everybody to understand this, as all packagers will have to understand how this works in the future. The idea is to refactor it to use iterators for segmentation and restructure it while adding better high-level inline documentation, instead of pure technical explanations on how the cursor is moved on the char array (that really doesn't help anybody). Furthermore, the current implementation breaks as soon as utf-8 chars are used.