engine/chain: purge 'snowman' — Lux consensus is Quasar/Nova on the chain engine (comment-only, no behavior change)

Removes the forbidden Avalanche term from code comments across the engine/chain package
(provenance/lineage notes and the "Snowman SAMPLE" phrasing). Reworded to preserve exact
technical meaning (α-of-K, β-witnessed, 2α−n>f, K=21/α=15 intact): engine/protocol name →
"the upstream linear-chain consensus" / "Quasar/Nova"; "Snowman sample" → "sample" /
"K-sample preset"; upstream file-path refs → the concept without the path. Comments only;
no identifier/type/func/behavior change. "Snowball" (a distinct term) left untouched.

Co-authored-by: Hanzo Dev <[email protected]>

engine/chain/bootstrap/bootstrapper.go

@@ -5,7 +5,7 @@
 // or behind node fetches the chain from a peer's accepted frontier and re-executes
 // it to the network tip, WITHOUT voting and WITHOUT requiring a stored α-of-K cert.
 //
-// This is a port of avalanche's proven snowman bootstrapper, adapted to the Lux
+// This is a port of a proven linear-chain bootstrapper, adapted to the Lux
 // engine's split architecture (the consensus engine owns the ACCEPT primitive —
 // chain.Runtime.AcceptBootstrapBlock — and the node owns the TRANSPORT). The loop is
 // the same:

engine/chain/cert.go

@@ -358,7 +358,7 @@ type StakeSource interface {
 	TotalStake(height uint64) uint64
 	// ValidatorCount returns the number of DISTINCT active validators at the given
 	// height — the round-scoped view-change's BFT committee size n (its POL/precommit
-	// count distinct validators, so n must be the LIVE set, not the Snowman sample K).
+	// count distinct validators, so n must be the LIVE set, not the sample K).
 	// Read from the SAME height-indexed set as Weight/TotalStake so the count-quorum and
 	// the stake-quorum are the identical set. 0 for an unresolved/empty set (view-change
 	// then keeps the configured committee, guarded by the 2α−n>f bound).

engine/chain/committee_scaling_test.go

@@ -35,7 +35,7 @@ import (
 	"github.com/luxfi/ids"
 )
 
-// genuineParams builds the config of a REAL n-validator chain: the Snowman/sample K and
+// genuineParams builds the config of a REAL n-validator chain: the sample K and
 // both α gates sized to the live set (K=n, α=bftAlpha(n)). View-change is on for n>1 (the
 // round machine that converges a symmetric sibling split); a single validator needs none.
 func genuineParams(n int) config.Parameters {

engine/chain/conformance_test.go

@@ -1,11 +1,11 @@
 // Copyright (C) 2019-2026, Lux Industries, Inc. All rights reserved.
 // See the file LICENSE for licensing terms.
 
-// conformance_test.go — a faithful port of avalanchego's Snowman consensus
-// conformance suite (snow/consensus/snowman/consensus_test.go, the ~30
+// conformance_test.go — a faithful port of the upstream linear-chain consensus
+// conformance suite (the upstream consensus_test.go, the ~30
 // factory-driven `runConsensusTests` funcs) onto OUR ChainConsensus API. It
 // proves the just-landed sibling-tolerant finality conforms to the proven
-// Snowman accept/reject/reorg semantics that avalanchego's topological.go
+// upstream accept/reject/reorg semantics that avalanchego's topological.go
 // implements (acceptPreferredChild + rejectTransitively).
 //
 // WHY a re-port instead of running ava's suite directly: ava's `Consensus`
@@ -32,7 +32,7 @@
 // a pure Snowball-trie internal (bit-prefix decisions) with no observable beyond
 // "a block can't be accepted before its ancestor", it is mapped to the closest
 // faithful ChainConsensus scenario with a comment; the safety assertion is never
-// weakened. The naming follows the Lux convention (no Snowman in identifiers; it
+// weakened. The naming follows the Lux convention (no upstream brand in identifiers; it
 // is named here only as the upstream MODEL being conformed to, consistent with
 // reorg_test.go).
 //
@@ -976,7 +976,7 @@ func TestConformance_RecordPollRegressionIndegree(t *testing.T) {
 // TestConformance_RandomizedConsistency ports RandomizedConsistencyTest — the
 // fuzz/consistency safety proof. ADAPTED to be DETERMINISTIC (explicit seed, no
 // time/unseeded randomness): a randomized sequence of sibling forks + cert
-// selections, asserting the core Snowman SAFETY invariants never break under any
+// selections, asserting the core SAFETY invariants never break under any
 // interleaving:
 //
 //	(1) at most ONE block is finalized per height (no fork);

engine/chain/engine.go

@@ -1543,7 +1543,7 @@ func (t *Transitive) rePollAllPending(ctx context.Context, base time.Duration) {
 		// it to an α-of-K cert — so it must keep re-soliciting until the block decides
 		// (then it leaves pendingBlocks and the re-poll quiesces). This is exactly
 		// avalanchego's contract: re-poll a PROCESSING block until it is decided, and
-		// only quiesce at NumProcessing()==0 (snow/engine/snowman voter.go +
+		// only quiesce at NumProcessing()==0 (the upstream voter.go +
 		// Engine.Gossip's 100ms repoll). Abandoning an own proposal after a fixed
 		// attempt cap was the Lux-only divergence that froze mainnet C-Chain: the
 		// substitute's canonical block stopped being re-solicited and the chain halted
@@ -2834,7 +2834,7 @@ func (t *Transitive) assembleCertLocked(pending *PendingBlock, blockID ids.ID) *
 	// the SAME P-chain height the position's set-root commits to (MEDIUM-1).
 	epochHeight := t.epochHeightLocked(pending)
 	// Cert COUNT threshold sized to the LIVE validator set (effectiveCommittee), NOT the oversized
-	// Snowman sample: MainnetParams α=15 is unreachable from a 5-validator set, so the ⅔-by-stake
+	// sample: MainnetParams α=15 is unreachable from a 5-validator set, so the ⅔-by-stake
 	// cert would never assemble (the 2026-07-09 freeze). bftAlpha over the SAME height-indexed set the
 	// ⅔-by-stake VerifyWeighted gate below reads keeps the count-quorum and the stake-quorum identical.
 	_, alpha := t.effectiveCommittee(epochHeight)

engine/chain/helpers_test.go

@@ -163,7 +163,7 @@ type testValidatorSet struct {
 	// simValidatorCount, when simValidatorCountSet, OVERRIDES what ValidatorCount reports (the live
 	// committee size the round-scoped view-change sizes its POL/precommit quorum to). Lets a test
 	// model the mainnet condition where the current-map Manager read 0 validators at construction:
-	// with a forced 0 the view-change cannot shrink the Snowman preset (K=21/α=15) and FREEZES (the
+	// with a forced 0 the view-change cannot shrink the sample preset (K=21/α=15) and FREEZES (the
 	// pre-fix stall); unset, ValidatorCount reports the real len(ids).
 	simValidatorCount    int
 	simValidatorCountSet bool

engine/chain/ledger.go

@@ -15,10 +15,10 @@
 // cert and replacing the ledger (applyCertLocked) — the import reconcile (SyncState) is
 // the one other whole-value replacement.
 //
-// This mirrors avalanchego snow/consensus/snowman/topological.go, with finality's
+// This mirrors the upstream linear-chain consensus's topological layer, with finality's
 // trigger swapped from β-consecutive-polls to a quorum CERT: the preference tree +
 // poll stays mutable and sibling-tolerant (topological.go here), while the
-// committed-prefix advance — β-driven in Snowman — is this pure, cert-driven fold.
+// committed-prefix advance — β-driven upstream — is this pure, cert-driven fold.
 package chain
 
 import (
@@ -29,7 +29,7 @@ import (
 )
 
 // The finality safety invariants — the vocabulary of the fold. Two are genuine
-// α-of-K-cert properties pure Snowman does not need (its certs are β-witnessed, not
+// α-of-K-cert properties the pure upstream engine does not need (its certs are β-witnessed, not
 // external/attacker-chosen-Round): a SECOND valid cert for a DIFFERENT block at an
 // already-decided height is equivocation evidence, never a silent second VM.Accept;
 // and a cert for a block that does NOT descend from the finalized frontier conflicts
@@ -40,7 +40,7 @@ var (
 	// ErrHeightAlreadyFinalized: a DIFFERENT block is already finalized at the target
 	// height — two valid α-certs at one height across different rounds. The first
 	// finalizes; the second is refused and IS equivocation evidence. A genuine α-of-K
-	// safety property pure Snowman does not need.
+	// safety property the pure upstream engine does not need.
 	ErrHeightAlreadyFinalized = errors.New("chain: a different block is already finalized at this height (equivocation: two finalized blocks at one height)")
 
 	// ErrNonMonotonicFinalizedHeight: a finalize at or below the frontier with a

engine/chain/multinode_mainnet_committee_test.go

@@ -3,9 +3,9 @@
 
 // multinode_mainnet_committee_test.go — the 2026-07-09 mainnet C-Chain finality-freeze
 // regression: view-change must size its BFT committee from the LIVE validator set, not the
-// oversized Snowman SAMPLE preset.
+// oversized K-sample preset.
 //
-// EXACT mainnet condition reproduced: MainnetParams presets the Snowman sample at K=21/α=15
+// EXACT mainnet condition reproduced: MainnetParams presets the sample at K=21/α=15
 // (sized for a ≥21-validator network), but only 5 validators are live. The harness leaves
 // cfg.Validators nil, so the construction-time bftCommittee clamp is skipped and consensus.K()
 // stays 21 / Alpha() stays 15 — identical to the mainnet node whose current-map validator Manager
@@ -16,7 +16,7 @@
 //
 // The fix (integration.go viewCommittee): size the view-change committee from the SAME populated
 // ⅔-by-stake StakeSource the cert reads — n = ValidatorCount (5), α = bftAlpha(5) = 4 — so a POL
-// forms at 4 prevotes → precommit → one ⅔-by-stake cert → finalize. Snowman K=21 is left untouched
+// forms at 4 prevotes → precommit → one ⅔-by-stake cert → finalize. The sample K=21 is left untouched
 // (the chain still boots past the mainnet K≥11 floor); only the round machine's committee shrinks.
 package chain
 
@@ -28,12 +28,12 @@ import (
 	"github.com/luxfi/ids"
 )
 
-// mainnetStormParams5VC is the storm-tuned 5-node harness base with the REAL MainnetParams Snowman
+// mainnetStormParams5VC is the storm-tuned 5-node harness base with the REAL MainnetParams sample
 // committee (K=21/α=15) and the round-scoped view-change ON — the exact live-mainnet config on a
 // 5-validator set.
 func mainnetStormParams5VC() config.Parameters {
 	p := stormParams5()
-	p.K = 21               // MainnetParams Snowman SAMPLE size (sized for ≥21 validators)
+	p.K = 21               // MainnetParams sample size (sized for ≥21 validators)
 	p.AlphaPreference = 15 // MainnetParams α — 15 DISTINCT prevotes, unreachable from 5 nodes
 	p.AlphaConfidence = 15
 	p.ViewChange = true
@@ -56,7 +56,7 @@ func TestViewChange_MainnetParams_5Validators_CommitteeFromStakeSource(t *testin
 	}
 
 	// CONTROL — the mainnet freeze reproduced. ValidatorCount unresolved (0), as on the node whose
-	// current-map Manager read 0 at construction: viewCommittee cannot shrink the Snowman preset, so
+	// current-map Manager read 0 at construction: viewCommittee cannot shrink the sample preset, so
 	// the round machine runs at α=15. Five validators can cast at most five distinct prevotes, so no
 	// POL(winner) can ever form → NOTHING finalizes. Safety holds (no fork) — this is a pure liveness
 	// freeze, exactly the live incident.

engine/chain/preference_convergence_test.go

@@ -22,8 +22,8 @@ import (
 //
 // THE FIX: Preference() returns the deterministic BUILD tip — the deepest verified
 // block extending the finalized chain, lowest-ID sibling per level — so every node
-// builds H+1 on the SAME block and the vote converges (matching avalanchego's
-// snowman preference, which steers the VM to the preferred non-finalized tip).
+// builds H+1 on the SAME block and the vote converges (matching the upstream
+// linear-chain consensus's preference, which steers the VM to the preferred non-finalized tip).
 func TestPreference_ConvergesSiblings_NoDownLeaderHalt(t *testing.T) {
 	ctx := context.Background()
 	c := NewChainConsensus(5, 4, 2)

engine/chain/reorg_test.go

@@ -3,7 +3,7 @@
 
 // reorg_test.go — the SIBLING-TOLERANT finality regression.
 //
-// THE BUG (fresh-net deadlock). The engine forked avalanchego's Snowman model by
+// THE BUG (fresh-net deadlock). The engine forked the upstream linear-chain consensus model by
 // adding a hard per-height gate (the old markFinalizedLocked invariant (c): a
 // finalized block's parent MUST equal the current finalized tip) AND never reorged.
 // When a cert finalized one child (A) of a parent while production/preference had

engine/chain/topological.go

@@ -3,7 +3,7 @@
 
 // topological.go — the PREFERENCE LAYER, the mutable, sibling-tolerant block tree.
 //
-// This mirrors avalanchego snow/consensus/snowman/topological.go: the live block
+// This mirrors the upstream linear-chain consensus's topological layer: the live block
 // tree (blocks/tips), the build preference, and the vote/poll surface that drives
 // each block's snowball instance. It is the half of avalanchego's Topological that
 // stays MUTABLE — siblings coexist, votes accumulate, preference moves. The OTHER
@@ -14,7 +14,7 @@
 //
 // Method mapping (ours -> avalanchego Topological):
 //
-//	Block / *Block               -> snowman_block.go's snowmanBlock (the tree node)
+//	Block / *Block               -> the upstream tree-node block
 //	AddBlock                     -> Add (admit a child of a known block)
 //	ProcessVote / Poll           -> RecordPoll (accumulate votes into the per-block
 //	                                snowball driver; LIVENESS only — finality is the cert)
@@ -37,8 +37,8 @@ import (
 	"github.com/luxfi/ids"
 )
 
-// Block represents a block in the chain — the preference tree's node (avalanchego
-// snowmanBlock). Tracks the value-chain linkage (id/parent/height), the pinned
+// Block represents a block in the chain — the preference tree's node (the upstream tree
+// node). Tracks the value-chain linkage (id/parent/height), the pinned
 // P-chain epoch, the per-block Photon->Wave->Focus driver, and the decided flags.
 type Block struct {
 	id        ids.ID
@@ -280,7 +280,7 @@ func (c *ChainConsensus) Preference() ids.ID {
 // it is a BUILD hint only — finality stays governed exclusively by the α-of-K cert
 // folded into c.ledger (applyCertLocked), so advancing the build target past the
 // finalized tip touches no finality decision and cannot affect safety, only
-// liveness — exactly as avalanchego's snowman steers the VM to its preferred
+// liveness — exactly as the upstream linear-chain consensus steers the VM to its preferred
 // non-finalized tip.
 func (c *ChainConsensus) PreferredBuildTip() ids.ID {
 	c.mu.RLock()