Roadmap Preview
2025Q3
- Webtransport (HTTP3)
- Continue implementation; may defer remaining work to Q4 depending on bandwidth.
- HTTP3 Finalized
- HTTP3 demands a complete working stack before anything functions - no partial implementation, this makes it non trivial and time consuming
- Gossipsub 2.0
- May continue depending on remaining gaps contingent on prior quarter’s outcomes.
- Kad-DHT
- Continue implementation; may defer remaining work to Q4 depending on bandwidth.
- Autonat V2
- Implement according to specs
- Maintenance
- Debug and fix flaky QA tests as well as continue giving support to QA team
- Remove go-libp2p-daemon usage in examples/tests
- (This is in addition to other maintenance tasks).
- Mix-core in nim-libp2p as Prototype / Unstable state
- Usable in PoCs or experiments
- Rough edges and minor issues expected, not recommended for production
- API and internal structure subject to changes
- Mix specs in libp2p/specs
- Upstream initial mix design and rationale to libp2p/spec as proposal or draft PR
2025Q4
- WebTransport (finalized)
- Complete implementation of WebTransport protocol
- Support WebTransport as a transport option within nim-libp2p
- Kad-DHT
- May continue depending on remaining gaps contingent on prior quarter’s outcomes.
- Cross-compilation for mobile
- Enable nim-libp2p builds for arm64/aarch64 targeting Android/iOS
- Needed for integration into Status App.
- C-bindings
- Expose stable C API surface for LogosCore and other external consumers.
- Connection abstraction refactor
- The current implementation of nim-libp2p includes abstractions that are difficult to understand, which makes development and collaboration more challenging. Refactoring the codebase would improve clarity, allowing contributors to grasp the code more easily. Additionally, a more streamlined implementation would enhance maintainability, leading to more efficient troubleshooting and debugging.
- IPv6 support
- Full handling of /ip6/… multiaddresses
- Mix-core in nim-libp2p as Done state
- core protocol logic stable;
- no blocking issues
- Issues (if any) are around API polish
- Maintenance
- evaluate whether to keep using nim-protobuf-serialization or migrate to minprotobuf
- (This is in addition to other maintenance tasks).
2026Q1 and beyond
- Unix Domain Socket transport (
/unix//tmp/foo) - Performance & scale
- Profiling and memory audit: guide future optimizations by identifying bottlenecks and alloc hotspots
- Implement zero-copy buffers between layes (transport / protocol / muxer / network libraries)
- Connection management
- enforce upper bounds on total open connections to prevent fd exhaustion, trim connections to a lower bound
- tag connections with metadata
- support protected connections (don’t trim them under pressure)
- implement exponential dial backoff to avoid peer hammering
- Modular Gossipsub
- Pluggable GossipSub scoring
- TODO
Uncategorized
The following items are not part of our current roadmap. They haven’t been requested by any consumer of nim-libp2p, so they do not have any prioritization status. Still, they are recorded here potential future consideration.
- bluetooth transport
- ble exploration by berty.
- Low-power proximity-based connections over BLE, mostly relevant in mobile/offline-first contexts.
- Not part of libp2p spec rn but has been explored in the ecosystem.
- tls secure transport
- spec.
- Reuse standard TLS instead of Noise for encryption/auth.
- webrtc-direct
- spec.
- Connects peers via WebRTC datachannels without signaling servers.
- Useful for browser-to-native or p2p-web setups.
- Available in js/go.
- mdns discovery
- spec
- local network peer discovery using multicast DNS.
- Simplifies peer bootstrapping in LANs.
- Trivial in theory, might depend on whether there are multicast DNS libraries in nim or having to implement one
- wasm support
- compile nim-libp2p to a wasm library for embedding in browser environments.
- no spec, but precedent exists in js-libp2p and rust efforts.
- native nim impl of Quic