I compared BencodeNET with quite some libraries to parse a whole lot of torrents. While mostly Bencode delivers the best results in some cases other libraries worked better when decoding paths and filenames. As it turns out MonoTorrent for example seems to prefer the "name.utf-8" and "path.utf-8" entries (if present!) over their non-"utf-8" counterparts.
Knowing that Bencode primarily wasn't meant as a torrent library: Is there any chance that this behaviour will be implemented in BencodeNET?