A go bencode serialization library could be smaller if it just serialized the bencode data to and from a generic dictionary container. I think another go BitTor

A go bencode serialization library could be smaller if it just serialized the bencode data to and from a generic dictionary container. I think another go BitTorrent client, gobit, takes this approach. But it's convenient to be able to serialize arbitrary application types. That makes the bencode library larger because of the need to use go's reflection APIs to analyze and access the application types. Go's reflection APIs are pretty verbose, which is why the line count is so high. To make things even more complicated, the BitTorrent protocol has a funny "compute-the-sha1-of-the-info-dictionary" requirement that forces BitTorrent clients to parse that particular part of the BitTorrent protocol using a generic parser. So in the end, the go bencode serializer supports both a generic dictionary parser and an application type parser, which makes it even larger.