Notes.md

• how do i prevent new connections?

• how to tell if a peer supports my protocol? (just attempt dial?)

• clients are both dialing kitsunet at eachother

• how will we ever get wrtc stable?

  • per-protocol connection limits

• stream each peer's networkState to cnc

• use peerBook for connections list libp2p.switch.muxedConn libp2p.getPeers (?)

• on connect, start timeout to check for kitsunet connection

• track pubsub

• opened libp2p/js-libp2p#175

• we want more low level hooks for rejecting connection attempts

• how will we ever get wrtc stable?

  • per-protocol connection limits

• network research

  • what to measure to determine network health?
  • what infrastructure to point at (instead of ipfs defaults)
  • how to stand up ws-star.cloud.ipfs.team
  • how do we limit peers to avoid crashing

• metamask light client

  • block sync

  • how to setup custom protocol/rpc over libp2p

  • pubsub with validation

  • how to shard data

  • how to process new txs to get new state

  • what telemetry is useful to gather

    • bandwidth
    • location
    • memory available?
    • connects/disconnects/discoveries
    • pubsub spanning tree
    • errors: sentry

  • @vyzo is working on pubsub

  • @JGAntunes is working on pubsub

    • ipfs/notes#266

  • platform deploy a "job" (eg peering strategy)

  • how to stand up ws-star.cloud.ipfs.team?

    • https://github.com/libp2p/js-libp2p-websocket-star-rendezvous

  • how do we limit peers to avoid crashing?

    • @pgte is working on this
    • application level prioritization of peers
    • ipfs-inactive/dynamic-data-and-capabilities#3 (comment)
    • watch repo https://github.com/libp2p/js-libp2p-connection-manager

  • how to configure the ipfs node for this experiment?

    • just boot libp2p?

  eth light client

  • GraphSync + selectors

  • implement custom protcols

    • dialProtocol
    • handleProtocol
    • example protocol - check dialer and listener inside src

    bridge (golang)

    • peer index
    • block syncer
    • state bridge
    • go-ipfs ipld selector alpha (handling) <--- daviddias
    • state prefetch
    • state transition (full)

    metamask (js) <--- daviddias

    • block tracking via ipns published head
    • ipld selector alpha (requesting)
    • broadcast new tx
    • log querying ????????
    • peering: bridge
    • block syncing

    light client perf hacks

    • eth_call proofs (eth_call as query)
    • pub/sub storage/logs on full node
    • extra-consensus geth bloom filter trie

    Better eth rpc

    • stream data (e.g. logs), with cancel (+ backpressure?)
    • selectors
    • query trace (anything sent message/eth to me)

    perf: ipld-resolver selector alpha peering: 1) bridge 2) metamask mesh new block + tx publishing coselector indexing tx -> block log querying, log -> tx