new features worth considering

[digizeph]

prefix-trie version 0.8 has added, a lot of the pure trie operations are already covered. Here is a list of related features that would be nice to consider.

Serialization / persistence:

• Zero-copy deserialization via rkyv or a flat binary format — useful for large routing tables (full DFZ is ~1M prefixes) where rebuild-from-iterator is too slow
• Streaming serde with framing (length-prefixed records) so you can append to a file or read from a socket without loading everything into memory

BGP / routing domain utilities:

• ROA validation — given a trie of ROA entries (prefix, max_length, origin_asn), validate whether a BGP announcement is RPKI-valid, invalid, or not-found
• Covering/covered prefix queries as first-class methods — ipnet-trie already has matches(), but you could add supernets_of(prefix) and subnets_of(prefix) with clearer semantics
• Prefix overlap detection — find all prefix pairs that overlap (useful for route leak analysis)
• Best-path simulation — store multiple values per prefix (e.g., multiple BGP routes) and select based on a comparator

Analytics:

• Prefix length distribution — histogram of /8 through /32 (v4) or /48 through /128 (v6) counts
• Address space coverage — what percentage of a given supernet is covered by entries in the trie
• Trie depth / density stats — useful for understanding routing table growth

Concurrency:

• A read-optimized concurrent wrapper (e.g., Arc<RwLock> with a nicer API, or a copy-on-write snapshot mechanism for lock-free reads)

Interop:

• Import/export in MRT TABLE_DUMP_V2 format (the standard BGP routing table dump format)
• Import from CSV/TSV (common for RPKI VRP feeds, IRR data)
• Conversion to/from ip_network_table (the other popular Rust IP lookup crate)