Cardano Updates

DeltaQ report

Signed-off-by: Chris Gianelloni <[email protected]>

Maps adversarial relay node infrastructure costs to measurable
front-running rates, with exploitable yield discounted by FIFO
batcher prevalence and DEX order fraction from script mapping.

Co-Authored-By: Claude Opus 4.6 <[email protected]>

Move ChainTree, ChainNode, and their 7 unit tests from consensus.rs
into chain_tree.rs. Add point() accessor so consensus doesn't reach
into ChainTree internals.

Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

Display tip_hash alongside tip_block_no in the inspector panel,
matching the format already used in topology nodes (e.g. "5 #a3b2").

Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

Emit NodeEvent::RolledBack for both peer-triggered rollbacks
(NetworkEvent::RolledBack) and consensus fork-switch rollbacks
(on_validation_complete returns true). UI shows violet flash on
topology nodes and magenta chip in event log.

Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

Track adopted_tip_hash separately from the chain tree's speculative
best tip. When on_validation_complete finds the new block is on a
different fork than the adopted tip, walk prev_hash links to find
the common ancestor and issue InjectRollback before InjectBlock.

Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

Replace the single local_tip in Consensus with a ChainTree that tracks
block headers keyed by hash, linked via prev_hash. Longest chain
(highest block_number) is selected as best tip. Blocks deeper than
security_param_k (default 2160) are pruned.

Block production now receives prev_hash and block_number from consensus
so headers contain correct chain linkage and globally consistent height.

Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

Use Typography component="div" so the Chip (renders <div>) isn't
nested inside a <p> element.

Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

Track blocks_produced in NodeSeriesPoint and display a green blocks
chart in the inspector, matching the aggregate charts style.

Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

Style recharts tooltips in aggregate charts and inspector panel with
dark backgrounds and white text to match the dashboard theme.

Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

Dark purple background, larger title left-aligned, node/edge count
on the right, white text throughout.

Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

Replace 1s HTTP polling with Server-Sent Events for event delivery.
Events now appear in the UI as they arrive from nodes, making block
propagation visible as a wave across the graph.

- Add GET /api/events/stream SSE endpoint with broadcast channel
- Add useEventStream hook with 50ms batching to avoid render storms
- Remove TipAdvanced from flash triggers (redundant with RBReceived)
- Protect "produced" flash from being overwritten by "received"

Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

Co-authored-by: Sebastian Nagel <[email protected]>

Cleanup

Replace the trivial [slot, hash] CBOR in fake blocks with proper
Shelley+ structure: era-tagged Conway header (10-field header_body with
slot, block_number, issuer_vkey, etc.) wrapped in #6.24, and a
complete block body with header + empty tx/witness/metadata maps.

This makes body.point() and WrappedHeader::parse() work correctly on
fake blocks, so the coordinator can derive the block point from the
body without needing the fetch range passed through.

Reverts the PeerEvent::BlockFetched { from, to, body } workaround
back to the original { body } form. The coordinator uses body.point()
again, which now works for both real and fake blocks. This is the
correct design for future multi-block range fetching.

New test: fake_block_has_parseable_cbor verifies header parsing,
point derivation from both header and body, and slot/era correctness.

324 tests (293 net-core + 31 net-node), 0 clippy warnings.

Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

Add telemetry system with event/stats sinks and per-peer bandwidth
monitoring:

net-core changes:
- mux/mod.rs: MuxStats with AtomicU64 bytes_sent/bytes_received counters,
  exposed via RunningMux
- mux/egress.rs, ingress.rs: increment counters per segment (header+payload)
- peer/types.rs: PeerEvent::Connected now carries Arc<MuxStats>
- multi_peer/types.rs: PeerInfo struct, NetworkCommand::QueryPeers,
  NetworkEvent::PeerSnapshot for per-peer telemetry snapshots
- coordinator.rs: stores mux_stats per peer, handles QueryPeers by
  building PeerInfo with address, mode, rtt, delay, bytes_sent/received

net-node changes:
- telemetry.rs: NodeEvent enum (sim-rs compatible JSONL format with
  serde tagged enum), StatsSnapshot with full peer list including
  bandwidth, EventSink/StatsSink traits, FileEventSink (JSONL),
  HttpEventSink (reqwest batched POST), LogStatsSink, HttpStatsSink,
  TelemetryHandle with counters and periodic flush
- config.rs: TelemetryConfig, EventSinkConfig (file/http),
  StatsSinkConfig (log/http)
- main.rs: stats timer sends QueryPeers, PeerSnapshot triggers
  emit_stats, all events recorded via telemetry.record()

323 tests (293 net-core + 30 net-node), 0 clippy warnings.
Live-tested: JSONL file output with correct sim-rs format, periodic
peer stats with bytes_sent/received incrementing, delay_ms reported.

Marks Stage 5 and all stages complete in test-node-plan.md.

Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

Extend the test node with Leios block types and network delay simulation:

- production.rs: EB/vote lotteries at stage boundaries, stage boundary
  detection, refactored to share lottery logic across block types
- consensus.rs: handles LeiosBlockOffered -> FetchLeiosBlock,
  LeiosVotesOffered -> FetchLeiosVotes, LeiosBlockTxsOffered ->
  FetchLeiosBlockTxs, logs Leios block/vote reception
- config.rs: ib/eb/vote generation probabilities, stage_length_slots,
  eb/vote validation timing, per-peer inbound_delay_ms
- Coordinator per-peer delay: new peer_delays config in CoordinatorConfig,
  events from delayed peers buffered and delivered after configured
  duration. Zero overhead when unconfigured (has_any_delays guard).
  Implemented in coordinator event loop with conditional select branch.

27 net-node tests + 293 net-core tests pass, 0 clippy warnings.
Live-tested: Leios EB/vote production and fetch, per-peer 200ms delay
visible in timestamps.

Marks Stage 4 complete in test-node-plan.md.

Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

The PeerCommand dispatch loop was duplicated nearly verbatim between
peer_task.rs and duplex_task.rs. Extract into command_dispatch.rs so
adding a new command variant requires a single edit instead of two.

Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

BlockBody::header() extracts the WrappedHeader directly from block body
CBOR, reusing shared try_extract_header() logic with point(). Coordinator
now falls back to body.header() when pending_headers misses, and skips
chain store insertion if neither produces a header. Post-remediation
review: added Resolution subsections for all three residual observations,
stripped trailing whitespace.

Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>