perf(fork-choice): early-break window walk in getCanonicalPayloadCounts

[lodekeeper]

Demo requested in #9815 (discussion_r3779437582) — showing how the O(window) early-break variant of getCanonicalPayloadCounts would look. Targets nflaig/circuit-breaker-canonical-counts, not unstable.

Approach

Instead of materializing the whole canonical chain via getAllAncestorNodes(), walk it newest-first and stop once a node drops below fromSlot. Ancestors are strictly slot-descending, so everything past that point is already out of the window — the previous full scan just continued over those.

const headIndex = this.getNodeIndexByRootAndStatus(headRoot, headPayloadStatus);
const head = headIndex !== undefined ? this.nodes[headIndex] : undefined;
if (head !== undefined && head.slot >= fromSlot) {
  countNode(head);
  for (const node of this.iterateAncestorNodesFromNode(head)) {
    if (node.slot < fromSlot) break;
    countNode(node);
  }
}

The head-inclusion gotcha (your note in r3779406069)

iterateAncestorNodes() starts one hop up (iterateAncestorNodesFromNode begins at .parent), so a direct swap drops the resolved head and undercounts by one. Handled here by counting the head explicitly first, then walking .parent via iterateAncestorNodesFromNode(head). A PENDING head is still skipped by countNode (same as getAllAncestorNodes not pushing a PENDING start node) — it only seeds the ancestor walk.

Behavior

Identical to the current full-scan: same canonical node sequence (head + ancestors), same filters (genesis / out-of-window / non-gloas / PENDING), same full/empty counts. Only the traversal is lazy with an early exit. The win (O(window) vs O(chain-to-anchor)) only shows under prolonged non-finality, when the chain-to-anchor grows well past the fault window.

Tests

Ran the fork-choice protoArray suite locally against this change — 131/131 pass (packages/fork-choice/test/unit/protoArray, 7 files), including every getCanonicalPayloadCounts case: genesis skip, "keeps EMPTY after a late FULL arrives", PENDING head, inclusive bounds, uses-supplied-head-branch. CI here re-runs the full suite.

Not attached to it — happy to close if you'd rather keep the simpler full-scan; just wanted to show the shape as you asked.

🤖 Generated with AI assistance

[ensi321]

This is good. Need this for #9815

[lodekeeper]

Thanks @ensi321 🙏

Since #9824 targets nflaig/circuit-breaker-canonical-counts (the #9815 branch), merging it drops the early-break straight into #9815 — it's mergeable as-is (no conflict with the metric-name refactor that landed on the branch since).

@nflaig — your call on the mechanism: merge #9824 onto the branch directly, or if you'd rather fold it into your own commit for authorship/history I'll close this out. Either works for me.

(Also retitled it — demo: isn't a valid conventional-commit type so the PR-title check was failing; it's perf(fork-choice): now.)

[ensi321]

I think while this is correct, I find it hard to follow with countNode

Something like

  const headIndex = this.getNodeIndexByRootAndStatus(headRoot, headPayloadStatus);
  let node = headIndex !== undefined ? this.nodes[headIndex] : undefined;

  while (node !== undefined && node.slot >= fromSlot) {
    if (
      node.slot !== GENESIS_SLOT &&
      node.slot <= toSlot &&
      isGloasBlock(node) &&
      node.payloadStatus !== PayloadStatus.PENDING
    ) 
    ...
    node = this.nodes[this.getParentNodeIndex(node)]
  }

would be more straightforward

[lodekeeper]

Good call — done in 95082e3 (pushed to the branch). Flattened to the single while loop walking getParentNodeIndex, dropping the countNode closure + iterateAncestorNodesFromNode generator. Reads much better.

Confirmed equivalent: the full protoArray unit suite is still green (131/131), including the tricky getCanonicalPayloadCounts cases — keeps-EMPTY-after-late-FULL, competing branches, genesis skip, PENDING head, inclusive bounds. The only mechanical difference is that this walks purely via getParentNodeIndex (no node.parent !== undefined pre-check), which coincides for the canonical in-window chain since the anchor's parent isn't in the array → getParentNodeIndex returns undefined and the loop ends.

[nflaig]

@lodekeeper is this ready to merge into my branch?

[lodekeeper]

Yeah, ready from my side 👍

• MERGEABLE, no conflict with the metric-name refactor on the branch
• protoArray unit suite green (131/131) — both after the initial commit and after ensi321's flatten (95082e3)
• The Validate PR title check passes; the UNSTABLE merge state is just because the full CI matrix doesn't run against a non-default base branch, not a failing check

Still your call on the mechanism — merge #9824 straight onto nflaig/circuit-breaker-canonical-counts, or if you'd rather fold it into your own commit for authorship/history I'll close this out. Either works for me.

[nflaig]

LGTM, this will be helpful during long non-finality

[lodekeeper]

🙏 thanks — and thanks @ensi321 for the review + the cleaner flat-loop suggestion.